<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	>

<channel>
	<title>data protection - KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</title>
	<atom:link href="https://www.kg-legal.eu/info/tag/data-protection/feed/" rel="self" type="application/rss+xml" />
	<link>https://www.kg-legal.eu/info/tag/data-protection/</link>
	<description>KIELTYKA GLADKOWSKI LEGAL &#124; CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</description>
	<lastBuildDate>Tue, 17 Nov 2026 19:48:14 +0000</lastBuildDate>
	<language>en-GB</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	
	<item>
		<title>Fighting Digital ID Fraud in 2026: From Industry Pilots to Statutory Trust Frameworks</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/fighting-digital-id-fraud-in-2026-from-industry-pilots-to-statutory-trust-frameworks/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/fighting-digital-id-fraud-in-2026-from-industry-pilots-to-statutory-trust-frameworks/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Tue, 17 Nov 2026 19:48:13 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[AI Act]]></category>
		<category><![CDATA[AMLR]]></category>
		<category><![CDATA[biometric verification]]></category>
		<category><![CDATA[Continuous Access Evaluation]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[data protection impact assessment]]></category>
		<category><![CDATA[deepfake fraud]]></category>
		<category><![CDATA[digital ID fraud]]></category>
		<category><![CDATA[digital identity fraud prevention]]></category>
		<category><![CDATA[digital identity in Poland]]></category>
		<category><![CDATA[digital identity law]]></category>
		<category><![CDATA[digital identity verification]]></category>
		<category><![CDATA[digital verification services]]></category>
		<category><![CDATA[eIDAS 2.0]]></category>
		<category><![CDATA[EUDI Wallet]]></category>
		<category><![CDATA[European Digital Identity Wallet]]></category>
		<category><![CDATA[fraud data sharing]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[identity fraud]]></category>
		<category><![CDATA[identity provider due diligence]]></category>
		<category><![CDATA[identity verification contracts]]></category>
		<category><![CDATA[injection attacks]]></category>
		<category><![CDATA[KG Legal]]></category>
		<category><![CDATA[KYC]]></category>
		<category><![CDATA[mObywatel Europa]]></category>
		<category><![CDATA[OfDIA]]></category>
		<category><![CDATA[OpenID Shared Signals Framework]]></category>
		<category><![CDATA[Payment Services Regulation]]></category>
		<category><![CDATA[PSD3]]></category>
		<category><![CDATA[reusable digital identity]]></category>
		<category><![CDATA[RISC]]></category>
		<category><![CDATA[shared signals]]></category>
		<category><![CDATA[social engineering]]></category>
		<category><![CDATA[UK digital identity trust framework]]></category>
		<category><![CDATA[UK GDPR]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=9042</guid>

					<description><![CDATA[<p>Publication date: November 17, 2026 From a 2023 pilot to a regulated market In July 2023 we attended a techUK webinar on fraud prevention in digital identity. Representatives of Synectics Solutions, Yoti and Mitek presented the &#8220;Shared Signals Pilot&#8221; (Project Shield): competing identity providers agreeing to exchange real-time alerts about compromised documents, credentials and identities. [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/fighting-digital-id-fraud-in-2026-from-industry-pilots-to-statutory-trust-frameworks/">Fighting Digital ID Fraud in 2026: From Industry Pilots to Statutory Trust Frameworks</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: November 17, 2026</mark></strong></p>



<h3 class="wp-block-heading"><a>From a 2023 pilot to a regulated market</a></h3>



<p>In July 2023 we attended a techUK webinar on fraud prevention in digital identity. Representatives of Synectics Solutions, Yoti and Mitek presented the &#8220;Shared Signals Pilot&#8221; (Project Shield): competing identity providers agreeing to exchange real-time alerts about compromised documents, credentials and identities. The panellists predicted that reusable digital identity would take off within three to five years, and that deepfakes would become the next major fraud vector.</p>



<span id="more-9042"></span>



<p>Three years on, both predictions have largely come true, but the landscape looks different from what anyone expected in 2023. Signal sharing has moved from a voluntary pilot to global technical standards and, in the EU payments sector, to a legal obligation. Digital identity verification in the UK now rests on a statutory framework. The EU is weeks away from its deadline for national digital identity wallets. And deepfakes are no longer a forecast but a measurable share of fraud attempts.</p>



<p>This article revisits the themes of that webinar against the legal and market position as of November 2026, with a focus on what it means for businesses operating in the UK, the EU and Poland.</p>



<h2 class="wp-block-heading"><a>The threat landscape in 2026</a></h2>



<p>The deepfake threat flagged by the webinar panellists is now mainstream. Entrust&#8217;s 2026 Identity Fraud Report, based on over 1 billion identity verifications across 195 countries, finds that deepfakes account for one in five biometric fraud attempts. Deepfaked selfie attempts rose by 58% in 2025, and injection attacks, which feed manipulated images or video directly into verification systems to bypass live capture, rose by 40% year on year.</p>



<p>Two other observations from 2023 still hold. First, different frauds need different responses: a teenager with a forged driving licence is not the same problem as an organised network reusing stolen identities at scale. Entrust describes fraud as now industrialised, running around the clock and offered &#8220;as a service&#8221;. Second, the human factor remains decisive. Social engineering and coercion, where victims are persuaded to use their own genuine credentials for the fraudster, are rising and cannot be stopped by document checks alone.</p>



<p>The practical consequence is that identity can no longer be treated as a one-off check at onboarding. It has to be monitored continuously, which is exactly where shared signals come in.</p>



<h2 class="wp-block-heading"><a>Shared signals: from pilot to global standard</a></h2>



<p>Project Shield was launched in June 2023 as a proof of concept. In Phase 1, Synectics&#8217; National SIRA platform, already used by banks, provided a dedicated pathway for Yoti and Mitek to create, send and receive signals about fake or fraudulently obtained identity documents, using test data, a real-time API and an alert system. The organisers openly acknowledged the main obstacles: privacy, data protection and the lack of a standardised framework for sharing.</p>



<p>The standardisation gap has since been closed at the technical level. In September 2025 the OpenID Foundation approved three Final Specifications: the Shared Signals Framework 1.0, the Continuous Access Evaluation Profile (CAEP) 1.0 and Risk Information Sharing and Coordination (RISC) 1.0. Together they define how one service can tell another, in real time, that a session was revoked, a credential changed or an account was compromised. Final status means the specifications are frozen and carry IP protections for implementers, which removes a key barrier to investment.</p>



<p>The legal question raised in 2023 is still the harder one. Sharing fraud intelligence means processing personal data about individuals who are suspected, not convicted, of wrongdoing. Under the GDPR and UK GDPR, any such scheme needs a clear legal basis, purpose limitation, retention limits and safeguards against wrongful flagging. As shown below, the EU has now chosen to legislate this basis directly for the payments sector.</p>



<h2 class="wp-block-heading"><a>United Kingdom: a statutory market, no national ID</a></h2>



<p>In 2023 the webinar speakers mentioned government requirements as one reason for their cooperation. Those requirements now have statutory force. Part 2 of the Data (Use and Access) Act 2025 (Royal Assent 19 June 2025) put the digital verification services (DVS) trust framework, a register of certified providers, a trust mark and an information gateway on a legal footing. Most of Part 2 came into force on 1 December 2025.</p>



<p>According to the Office for Digital Identities and Attributes (OfDIA) annual report of 14 July 2026:</p>



<ul class="wp-block-list">
<li>the DVS register listed 46 providers offering 64 certified services;</li>



<li>version 1.0 of the trust framework, the first statutory version, was published on 9 June 2026 and takes effect from 1 September 2026;</li>



<li>the &#8220;UK CertifID&#8221; trust mark was designated in March 2026 for certified, registered providers;</li>



<li>the information gateway allowing public authorities to share data with registered providers at an individual&#8217;s request awaits a Code of Practice;</li>



<li>OfDIA is reviewing the security and fraud sections of the trust framework.</li>
</ul>



<p>The public-sector picture has moved in the opposite direction. In September 2025 the government announced a national digital ID that would become mandatory for right-to-work checks. In January 2026 the mandatory element was dropped, and a public consultation ran from March to May 2026. In July 2026 the new Prime Minister, Andy Burnham, scrapped the scheme altogether. GOV.UK One Login, the government&#8217;s existing sign-in and verification service, continues to expand and had been joined by 245 services by October 2026.</p>



<p>For businesses, the result is a market led by certified private providers operating under a statutory framework, rather than a single state-issued credential. Certification and registration are becoming the practical benchmark of trustworthiness for UK identity checks.</p>



<h2 class="wp-block-heading"><a>European Union: wallets, mandatory fraud data sharing and deepfake labelling</a></h2>



<p>The EU has taken a more regulatory route, with four instruments converging on digital identity fraud.</p>



<p><strong>eIDAS 2.0 and the EU Digital Identity Wallet.</strong> Under Regulation (EU) 2024/1183, every Member State must make at least one EUDI Wallet available by 24 December 2026. Use is voluntary for individuals. From December 2027, regulated private relying parties, including banks, insurers, telecoms operators and very large online platforms, will have to accept the wallet when a user chooses to use it. Readiness is uneven: a Q1 2026 industry review rated only three Member States as almost certain to meet the deadline and four as unlikely. The wallet is designed for selective disclosure, so a user can prove a single attribute, such as being over 18, without handing over a full document. This is the &#8220;reusable ID&#8221; the 2023 panellists described, now with a legal mandate.</p>



<p><strong>The Payment Services Regulation (PSR).</strong> The final compromise texts of PSD3 and the PSR were published in April 2026, and the Council adopted its first-reading position on 29 October 2026. Publication in the Official Journal is still pending, with most rules expected to apply about 21 months after entry into force, i.e. in 2028. The PSR turns the Project Shield idea into a legal obligation: payment service providers must share fraud-related data with each other through dedicated platforms, with time-limited retention and an express legal basis under the GDPR. It also extends providers&#8217; liability for certain impersonation fraud where fraud-prevention measures are inadequate.</p>



<p><strong>The AI Act.</strong> The transparency obligations in Article 50 of Regulation (EU) 2024/1689 have applied since 2 August 2026, including the duty of deployers to disclose that image, audio or video content is a deepfake. The Digital Omnibus (Regulation (EU) 2026/1744) postponed the high-risk regime, but left Article 50 largely intact, granting only a grace period until 2 December 2026 for machine-readable marking by generative AI systems already on the market. Labelling will not stop fraudsters, who will simply ignore it, but it raises the compliance bar for legitimate providers of synthetic media tools.</p>



<p><strong>The Anti-Money Laundering Regulation (AMLR).</strong> Its key obligations apply from 10 July 2027 and expressly allow electronic identification under eIDAS, including the EUDI Wallet, for customer due diligence. For obliged entities, this links wallet acceptance directly to KYC processes.</p>



<h2 class="wp-block-heading"><a>Poland: mObywatel Europa</a></h2>



<p>Poland enters the wallet era from a strong position. The mObywatel app already has nearly 10 million users on version 2.0. In February 2026 reports suggested that the existing app did not meet the EU technical requirements and that a separate wallet app would be needed. In March 2026 the Ministry of Digital Affairs confirmed instead that the EUDI Wallet will be integrated into mObywatel, with no second app for users.</p>



<p>The wallet is being delivered by the Central Informatics Centre (COI) under the name mObywatel Europa. On 16 September 2026 the Ministry opened a free testing sandbox where businesses and institutions can integrate their services with a demonstration build. A pilot launch is planned for December 2026, while full integration may take until 2027 or 2028. Poland has also taken part in the EU large-scale pilots POTENTIAL and APTITUDE.</p>



<p>For Polish banks, payment institutions, insurers and telecoms operators, the sandbox is the practical starting point. The acceptance obligation from December 2027 will coincide closely with the AMLR (July 2027) and, later, the PSR, so integration work is best planned as a single programme rather than three separate projects.</p>



<h2 class="wp-block-heading"><a>What businesses should do now</a></h2>



<p>The central message of the 2023 webinar, that no single organisation can fight identity fraud alone, has been confirmed by both the market and the legislator. In practice, we recommend that businesses relying on remote identity checks:</p>



<ol class="wp-block-list">
<li><strong>Map their obligations by jurisdiction and date.</strong> EUDI Wallet acceptance (December 2027), AMLR customer due diligence (July 2027), AI Act Article 50 (in force since August 2026) and the PSR (expected 2028) overlap for many financial and digital businesses.</li>



<li><strong>Test early.</strong> Polish entities can use the mObywatel Europa sandbox; UK-facing businesses should check that their identity providers are certified and listed on the DVS register.</li>



<li><strong>Review vendor contracts.</strong> Agreements with identity verification providers should address deepfake and injection-attack detection, certification status, incident notification and allocation of liability for fraud losses.</li>



<li><strong>Prepare a legal basis for fraud data sharing.</strong> Before joining any signal-sharing scheme, document the legal basis, carry out a data protection impact assessment, set retention limits and provide a route for individuals to challenge a wrongful flag.</li>



<li><strong>Treat identity as continuous.</strong> Onboarding checks should be complemented by session-level and account-level signals, in line with the OpenID Shared Signals standards.</li>
</ol>



<p>As the panellists said in 2023, this requires patience and &#8220;baby steps&#8221;. The difference in 2026 is that many of those steps now come with statutory deadlines.</p>



<h2 class="wp-block-heading"><a>Sources</a></h2>



<p>techUK, <a href="https://www.techuk.org/what-we-deliver/events/fraud-prevention-in-digital-id.html" target="_blank" rel="noreferrer noopener">Fraud prevention in digital ID</a> (webinar, 20 July 2023)</p>



<p>Yoti, <a href="https://yoti.com/blog/digital-id-shared-signals-pilot-launched-to-tackle-evolving-fraud-risks" target="_blank" rel="noreferrer noopener">Digital ID Shared Signals Pilot launched to tackle evolving fraud risks</a> (2023)</p>



<p>Biometric Update, <a href="https://www.biometricupdate.com/202306/yoti-mitek-and-synectics-launch-pilot-to-fight-digital-id-fraud-with-shared-signals" target="_blank" rel="noreferrer noopener">Yoti, Mitek and Synectics launch pilot to fight digital ID fraud with shared signals</a></p>



<p>OpenID Foundation, <a href="https://openid.net/press-release-openid-foundation-finalizes-global-standards-for-real-time-identity-security/" target="_blank" rel="noreferrer noopener">Finalizes global standards for real-time identity security</a> (16 September 2025)</p>



<p>Entrust, <a href="https://www.entrust.com/resources/reports/identity-fraud-report" target="_blank" rel="noreferrer noopener">2026 Identity Fraud Report</a></p>



<p>GOV.UK, <a href="https://www.gov.uk/government/publications/office-for-digital-identities-and-attributes-2026-annual-report/ofdia-2026-annual-report-on-the-operation-of-part-2-of-data-use-and-access-act-2025" target="_blank" rel="noreferrer noopener">OfDIA 2026 Annual Report on Part 2 of the Data (Use and Access) Act 2025</a> (14 July 2026)</p>



<p>GOV.UK blog, <a href="https://enablingdigitalidentity.blog.gov.uk/2025/12/01/dvs-measures-in-the-data-act-have-come-into-force/" target="_blank" rel="noreferrer noopener">DVS measures in the Data Act have come into force</a> (1 December 2025)</p>



<p>Lewis Silkin, <a href="https://www.lewissilkin.com/en/insights/2026/03/23/uk-government-opens-consultation-on-digital-id-scheme-102mnlk" target="_blank" rel="noreferrer noopener">UK Government opens consultation on digital ID scheme</a> (March 2026)</p>



<p>Computer Weekly, <a href="https://computerweekly.com/news/366646014/Burnham-government-to-drop-digital-ID-and-tech-department" target="_blank" rel="noreferrer noopener">Burnham government to drop digital ID</a> (July 2026)</p>



<p>Tech Insider, <a href="https://tech-insider.org/uk-digital-id-scrapped-one-login-752-million-2026/" target="_blank" rel="noreferrer noopener">UK digital ID scrapped, One Login continues</a> (October 2026)</p>



<p>Namirial, <a href="https://www.namirial.com/en/blog/stories/status-check-eudi-wallet/" target="_blank" rel="noreferrer noopener">Status check: 9 months to the EUDI Wallet</a> (Q1 2026)</p>



<p>Payment Expert, <a href="https://paymentexpert.com/2026/11/10/psd3-guide-eu-payment-services/" target="_blank" rel="noreferrer noopener">PSD3 guide</a> (10 November 2026)</p>



<p>Worldline, <a href="https://worldline.com/en/home/main-navigation/resources/blogs/2026/the-scope-and-timeline-are-locked-in-for-psd3-and-psr-what-should-psps-know" target="_blank" rel="noreferrer noopener">The scope and timeline are locked in for PSD3 and PSR</a></p>



<p>IAPP, <a href="https://iapp.org/news/a/new-eu-payment-rules-could-expand-fraud-monitoring" target="_blank" rel="noreferrer noopener">New EU payment rules could expand fraud monitoring</a> (April 2026)</p>



<p>Ropes &amp; Gray, <a href="https://www.ropesgray.com/en/insights/viewpoints/2026/08/102nfqm/you-talkin-to-me-operationalising-the-eu-ai-acts-transparency-obligations" target="_blank" rel="noreferrer noopener">Operationalising the EU AI Act&#8217;s transparency obligations</a> (August 2026)</p>



<p>Reed Smith, <a href="https://www.reedsmith.com/our-insights/blogs/viewpoints/102nfqg/always-read-the-label-eu-ai-transparency-rules-and-questions-that-remain-unanswe/" target="_blank" rel="noreferrer noopener">EU AI transparency rules and questions that remain unanswered</a></p>



<p>bank.pl, <a href="https://bank.pl/?p=559565" target="_blank" rel="noreferrer noopener">IDENTT dołączył do środowiska testowego mObywatel Europa</a> (2026)</p>



<p>Telepolis, <a href="https://www.telepolis.pl/tech/aplikacje/mobywatel-europejski-portfel-tozsamosci-cyfrowej" target="_blank" rel="noreferrer noopener">mObywatel Europa od grudnia</a></p>



<p>iGrant.io, <a href="https://eudi-wallet-tracker.igrant.io/tracker/poland/" target="_blank" rel="noopener">EUDI Wallet tracker: Poland</a></p>
<p> </p>



<p>Services</p>



<p><strong>KIELTYKA GLADKOWSKI</strong> is a Polish business law firm headquartered in Warsaw and Krakow, with a New York office under the iSTART1 brand. Building on a practice dating back to 1981, we advise clients in the following areas.</p>



<h1 class="wp-block-heading">Explore our services</h1>



<p>Select a practice area to learn more.</p>



<figure class="wp-block-table"><table class="has-luminous-vivid-amber-background-color has-background has-fixed-layout"><thead><tr><td><strong>Business<br>&amp; Finance</strong></td><td><strong>Technology<br>&amp; IP</strong></td><td><strong>Energy<br>&amp; Projects</strong></td><td><strong>Sectors &amp;<br>Cross Border</strong></td><td><strong>Private Clients<br>&amp; Disputes</strong></td></tr></thead><tbody><tr><td><a href="https://www.kg-legal.pl/investment-law-and-processes-in-poland-1/">Commercial &amp; Corporate</a> <a href="https://www.kg-legal.eu/polish-limited-liability-company-legal-support-in-operational-business-maintenance-and-representation-of-shareholders-in-court/">Polish LLCs</a> <a href="https://www.kg-legal.eu/employment-law-and-collective-labour-law-in-poland-including-employment-contracts/">Employment Law</a></td><td><a href="https://www.kg-legal.pl/it-bio-it-new-technologies-media-law-1/">IT, IP, Media &amp; New Tech</a> <a href="https://www.kg-legal.eu/trademarks-and-brand-protection/">Trademarks &amp; Brand Protection</a></td><td><a href="https://www.kg-legal.eu/green-transition/">Green Transition</a> <a href="https://www.kg-legal.eu/energy-transformation-in-poland-main-assumptions-of-the-polish-energy-policy-2040-2/">Energy Policy 2040</a></td><td><a href="https://www.kg-legal.pl/pharmaceutical-life-sciences-healthcare-law-1/">Pharma, Life Sciences &amp; Healthcare</a> <a href="https://www.kg-legal.eu/pharmaceutical-law/">Pharmaceutical Law</a></td><td><a href="https://www.kg-legal.eu/private-client-desk/">Private Client Desk</a> <a href="https://www.kg-legal.eu/hnw-and-next-gen-wealth/">HNW &amp; Next Gen Wealth</a></td></tr><tr><td><a href="https://www.kg-legal.eu/venture-capital-and-private-equity/">Venture Capital &amp; Private Equity</a> <a href="https://www.kg-legal.eu/stock-exchange-and-securities-law-in-poland/">Stock Exchange &amp; Securities</a> <a href="https://www.kg-legal.eu/markets-in-modern-and-conventional-financial-instruments-including-tokens-crypto-assets-packaged-retail-and-insurance-based-investment-products/">Financial Instruments &amp; Crypto</a></td><td><a href="https://www.kg-legal.eu/artificial-intelligence-ai-law-poland-eu/">AI Sector</a> <a href="https://www.kg-legal.eu/data-centers/">Data Centers</a></td><td><a href="https://www.kg-legal.eu/regulatory-law-of-the-energy-sector-in-poland/">Energy Sector</a> <a href="https://www.kg-legal.eu/green-energy-law-renewable-energy-law/">Renewable Energy</a></td><td><a href="https://www.kg-legal.eu/medical-devices-including-medical-devices-law-under-combined-atmps/">Medical Devices</a> <a href="https://www.kg-legal.eu/food-law/">Food Law</a></td><td><a href="https://www.kg-legal.eu/international-divorces/">International Divorces</a> <a href="https://www.kg-legal.eu/international-family-law/">International Family Law</a></td></tr><tr><td><a href="https://www.kg-legal.eu/agreements-and-contracts/">Agreements &amp; Contracts</a> <a href="https://www.kg-legal.eu/contracts-in-the-international-trade-smart-contracts/">International Trade &amp; Smart Contracts</a></td><td><a href="https://www.kg-legal.eu/data-privacy-and-data-protection/">Data Privacy &amp; Protection</a> <a href="https://www.kg-legal.pl/cybersecurity/">Cybersecurity</a></td><td><a href="https://www.kg-legal.eu/biofuels-and-biocomponents/">Biofuels &amp; Biocomponents</a> <a href="https://www.kg-legal.eu/offshore-wind-farms/">Offshore Wind</a></td><td><a href="https://www.kg-legal.eu/art-market/">Art Market</a> <a href="https://www.kg-legal.eu/legal-services-in-fashion-law/">Fashion</a></td><td><a href="https://www.kg-legal.eu/cross-border-inheritance-matters/">Cross Border Inheritance</a> <a href="https://www.kg-legal.eu/immigration-law-business-immigration-polish-passport/">Immigration &amp; Polish Passports</a></td></tr><tr><td><a href="https://www.kg-legal.eu/offshore-companies-tax-optimization/">Offshore Companies &amp; Tax</a></td><td><a href="https://www.kg-legal.eu/distributed-ledger-technology-law/">Distributed Ledger Technology</a> <a href="https://www.kg-legal.eu/metaverse-and-web3-legal-environment/">Metaverse &amp; Web3</a></td><td><a href="https://www.kg-legal.eu/greenfield-in-poland/">Greenfield Investment</a> <a href="https://www.kg-legal.eu/kieltyka-gladkowski-provides-assistance-in-the-area-of-european-partnerships-in-horizon-europe-programme/">Horizon Europe</a></td><td><a href="https://www.kg-legal.pl/cross-border-cases-1/">Cross Border Practice</a> <a href="https://www.kg-legal.eu/customs-and-tariffs-support-for-international-traders-in-poland-and-eu/">Customs &amp; Tariffs</a></td><td><a href="https://www.kg-legal.eu/asset-tracing-in-poland-for-international-creditors/">Asset Tracing in Poland</a> <a href="https://www.kg-legal.eu/asset-tracing/">Asset Tracing</a> <a href="https://www.kg-legal.eu/business-investigations-private-detective-services/">Business Investigations</a></td></tr><tr><td><a href="https://www.kg-legal.eu/dissolution-and-liquidation-of-capital-companies/">Company Dissolution &amp; Liquidation</a> <a href="https://www.kg-legal.eu/cross-border-bankruptcy-in-poland/">Cross Border Bankruptcy</a></td><td><a href="https://www.kg-legal.eu/dual-use-technology-law/">Dual Use Technology</a> <a href="https://www.kg-legal.eu/drone-uas-law/">Drones &amp; UAS</a> <a href="https://www.kg-legal.eu/defence-and-space-sector/">Defence &amp; Space</a></td><td><a href="https://www.kg-legal.eu/industrial-micro-and-public-macro-infrastructure/">Infrastructure</a> <a href="https://www.kg-legal.eu/construction/">Construction</a> <a href="https://www.kg-legal.eu/public-private-partnerships/">Public Private Partnerships</a></td><td><a href="https://www.kg-legal.eu/sworn-certified-legal-translations-of-documents-and-contracts/">Sworn &amp; Certified Legal Translations</a></td><td><a href="https://www.kg-legal.eu/litigation-and-representation-of-international-clients-before-the-polish-courts/">Litigation in Polish Courts</a> <a href="https://www.kg-legal.eu/litigation-antitrust-and-competition-law/">Litigation &amp; Competition</a></td></tr></tbody></table></figure>



<p>Service directory: <a href="https://www.kg-legal.eu/services/">kg-legal.eu/services</a></p>
<p> </p>




<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/fighting-digital-id-fraud-in-2026-from-industry-pilots-to-statutory-trust-frameworks/">Fighting Digital ID Fraud in 2026: From Industry Pilots to Statutory Trust Frameworks</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/fighting-digital-id-fraud-in-2026-from-industry-pilots-to-statutory-trust-frameworks/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Thuringia Meets Małopolska: KG Legal KIELTYKA GLADKOWSKI at the Business Meeting with Minister-President Mario Voigt at SOLARIS</title>
		<link>https://www.kg-legal.eu/info/investment-law-and-processes-in-poland/thuringia-meets-malopolska-kg-legal-kieltyka-gladkowski-at-the-business-meeting-with-minister-president-mario-voigt-at-solaris/</link>
					<comments>https://www.kg-legal.eu/info/investment-law-and-processes-in-poland/thuringia-meets-malopolska-kg-legal-kieltyka-gladkowski-at-the-business-meeting-with-minister-president-mario-voigt-at-solaris/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Tue, 17 Nov 2026 13:43:41 +0000</pubDate>
				<category><![CDATA[INVESTMENT LAW AND PROCESSES IN POLAND]]></category>
		<category><![CDATA[advanced manufacturing]]></category>
		<category><![CDATA[AHK Polska]]></category>
		<category><![CDATA[AI Act]]></category>
		<category><![CDATA[Artificial intelligence]]></category>
		<category><![CDATA[biotechnology]]></category>
		<category><![CDATA[Business in Małopolska]]></category>
		<category><![CDATA[Cross-Border Legal Services]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[Doing business in Poland]]></category>
		<category><![CDATA[Gaia AI Factory]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[German business in Poland]]></category>
		<category><![CDATA[German investment in Poland]]></category>
		<category><![CDATA[industrial automation]]></category>
		<category><![CDATA[Intellectual Property]]></category>
		<category><![CDATA[KG Legal Kieltyka Gladkowski]]></category>
		<category><![CDATA[Kraków]]></category>
		<category><![CDATA[LEG Thüringen]]></category>
		<category><![CDATA[Life Sciences]]></category>
		<category><![CDATA[Mario Voigt]]></category>
		<category><![CDATA[market entry in Poland]]></category>
		<category><![CDATA[Małopolska]]></category>
		<category><![CDATA[mechatronics]]></category>
		<category><![CDATA[Medical devices]]></category>
		<category><![CDATA[MedTech]]></category>
		<category><![CDATA[microelectronics]]></category>
		<category><![CDATA[Polish-German business cooperation]]></category>
		<category><![CDATA[R&D cooperation]]></category>
		<category><![CDATA[research agreements]]></category>
		<category><![CDATA[Research and Development]]></category>
		<category><![CDATA[Research Commercialisation]]></category>
		<category><![CDATA[SOLARIS]]></category>
		<category><![CDATA[technology partnerships]]></category>
		<category><![CDATA[Technology Transfer]]></category>
		<category><![CDATA[Thuringia]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=9035</guid>

					<description><![CDATA[<p>Publication date: November 17, 2026 Kraków, 17 November 2026 Today KG Legal is taking part in a business meeting with a delegation from the Free State of Thuringia, led by Minister-President Mario Voigt, hosted at the National Synchrotron Radiation Centre SOLARIS in Kraków. The meeting is organised by the Polish-German Chamber of Industry and Commerce [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/investment-law-and-processes-in-poland/thuringia-meets-malopolska-kg-legal-kieltyka-gladkowski-at-the-business-meeting-with-minister-president-mario-voigt-at-solaris/">Thuringia Meets Małopolska: KG Legal KIELTYKA GLADKOWSKI at the Business Meeting with Minister-President Mario Voigt at SOLARIS</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: November 17, 2026</mark></strong></p>



<p><em>Kraków, 17 November 2026</em></p>



<figure class="wp-block-image size-large"><img fetchpriority="high" decoding="async" width="1024" height="538" src="https://www.kg-legal.eu/wp-content/uploads/2026/11/Blog-header-1200×630@1x-1024x538.png" alt="" class="wp-image-9036" srcset="https://www.kg-legal.eu/wp-content/uploads/2026/11/Blog-header-1200×630@1x-1024x538.png 1024w, https://www.kg-legal.eu/wp-content/uploads/2026/11/Blog-header-1200×630@1x-300x158.png 300w, https://www.kg-legal.eu/wp-content/uploads/2026/11/Blog-header-1200×630@1x-768x403.png 768w, https://www.kg-legal.eu/wp-content/uploads/2026/11/Blog-header-1200×630@1x.png 1200w" sizes="(max-width: 1024px) 100vw, 1024px" /></figure>



<p>Today KG Legal is taking part in a business meeting with a delegation from the Free State of Thuringia, led by Minister-President Mario Voigt, hosted at the National Synchrotron Radiation Centre SOLARIS in Kraków. The meeting is organised by the Polish-German Chamber of Industry and Commerce (AHK Polska) together with the Marshal&#8217;s Office of the Małopolska Region and the Business in Małopolska centre, with LEG Thüringen, the Thuringian economic development agency, as a partner.</p>



<span id="more-9035"></span>



<p>The aim is simple and ambitious at the same time: to build direct business relationships between companies, research institutes and institutions from Thuringia and Małopolska, and to open new avenues for cooperation in industry, technology and R&amp;D.</p>



<h2 class="wp-block-heading">A venue that sets the tone</h2>



<p>It is hard to imagine a better setting for a meeting about science-driven business than SOLARIS, Poland&#8217;s synchrotron facility in Kraków. The programme combines an official opening with representatives of both regions and SOLARIS, presentations on the economic and technological potential of Małopolska and Thuringia, a presentation of the Polish AI factory Gaia AI Factory, and a guided tour of the synchrotron. The afternoon closes with short company profiles from the German participants and a networking session dedicated to one-to-one conversations.</p>



<h2 class="wp-block-heading">Who came from Thuringia</h2>



<p>The Thuringian delegation reflects the region&#8217;s strengths in high technology and advanced manufacturing. It includes medtech companies such as 3di GmbH (patient-specific implants and 3D surgical models) and HEMA Orthopädische Systeme GmbH (orthopaedic solutions and medical technology), leading research organisations such as Fraunhofer IOF in Jena (photonics and optical systems), the DLR Institute for Data Science in Jena (AI and data analysis), IMMS GmbH (microelectronics, mechatronics, sensor systems and AI) and iba e.V. (bioprocess and analytical technologies), as well as industrial companies GSM GmbH and GSN Maschinen-Anlagen-Service GmbH, the certification body TÜV Thüringen e.V. and LEG Thüringen.</p>



<p>The focus sectors &#8211; microelectronics, mechatronics, industrial automation, biotechnology, life sciences and medtech &#8211; overlap closely with Małopolska&#8217;s own innovation profile, including its life sciences cluster, universities and research infrastructure.</p>



<h2 class="wp-block-heading">Where law meets technology</h2>



<p>For KG Legal, these are familiar sectors. Life sciences, healthcare, medtech, AI and data protection are at the core of our practice, and we regularly advise foreign companies, including German businesses, entering or expanding in Poland. Cross-border technology cooperation tends to raise a recurring set of legal questions, and it pays to address them early:</p>



<ul class="wp-block-list">
<li><strong>Medical devices and life sciences products</strong> &#8211; placing products on the Polish market, Polish-language labelling and instructions for use, obligations towards the Polish regulator (URPL), distribution models and public procurement for hospitals.</li>



<li><strong>R&amp;D partnerships with universities and research institutes</strong> &#8211; ownership and licensing of intellectual property, publication rights, commercialisation of research results and spin-out structures, as well as EU and national grant funding.</li>



<li><strong>AI and data</strong> &#8211; compliance with the EU AI Act, GDPR and the specific rules on health data, including in joint research projects.</li>



<li><strong>Market entry and operations</strong> &#8211; company set-up, contracts with Polish partners and distributors, employment, and effective dispute resolution across borders.</li>
</ul>



<p>Poland and Germany share a single market and a long history of economic ties, but successful cooperation still depends on getting the details right on both sides of the border.</p>



<h2 class="wp-block-heading">Looking ahead</h2>



<p>We thank AHK Polska, the Marshal&#8217;s Office of the Małopolska Region, Business in Małopolska, LEG Thüringen and NCPS SOLARIS for bringing the two regions together. Meetings like this one show how much potential lies in Polish-German cooperation in science and technology, and we look forward to following up on today&#8217;s conversations.</p>



<p>If your organisation is considering projects in Poland, or cooperation with partners in Małopolska, our team in Kraków and Warsaw will be glad to help. You can reach us at office@kg-legal.pl.</p>
<p> </p>



<h2 class="wp-block-heading">Services</h2>



<p><strong>KIELTYKA GLADKOWSKI</strong> is a Polish business law firm headquartered in Warsaw and Krakow, with a New York office under the iSTART1 brand. Building on a practice dating back to 1981, we advise clients in the following areas.</p>



<h1 class="wp-block-heading">Explore our services</h1>



<p>Select a practice area to learn more.</p>



<figure class="wp-block-table"><table class="has-luminous-vivid-amber-background-color has-background has-fixed-layout"><thead><tr><td><strong>Business<br>&amp; Finance</strong></td><td><strong>Technology<br>&amp; IP</strong></td><td><strong>Energy<br>&amp; Projects</strong></td><td><strong>Sectors &amp;<br>Cross Border</strong></td><td><strong>Private Clients<br>&amp; Disputes</strong></td></tr></thead><tbody><tr><td><a href="https://www.kg-legal.pl/investment-law-and-processes-in-poland-1/">Commercial &amp; Corporate</a> <a href="https://www.kg-legal.eu/polish-limited-liability-company-legal-support-in-operational-business-maintenance-and-representation-of-shareholders-in-court/">Polish LLCs</a> <a href="https://www.kg-legal.eu/employment-law-and-collective-labour-law-in-poland-including-employment-contracts/">Employment Law</a></td><td><a href="https://www.kg-legal.pl/it-bio-it-new-technologies-media-law-1/">IT, IP, Media &amp; New Tech</a> <a href="https://www.kg-legal.eu/trademarks-and-brand-protection/">Trademarks &amp; Brand Protection</a></td><td><a href="https://www.kg-legal.eu/green-transition/">Green Transition</a> <a href="https://www.kg-legal.eu/energy-transformation-in-poland-main-assumptions-of-the-polish-energy-policy-2040-2/">Energy Policy 2040</a></td><td><a href="https://www.kg-legal.pl/pharmaceutical-life-sciences-healthcare-law-1/">Pharma, Life Sciences &amp; Healthcare</a> <a href="https://www.kg-legal.eu/pharmaceutical-law/">Pharmaceutical Law</a></td><td><a href="https://www.kg-legal.eu/private-client-desk/">Private Client Desk</a> <a href="https://www.kg-legal.eu/hnw-and-next-gen-wealth/">HNW &amp; Next Gen Wealth</a></td></tr><tr><td><a href="https://www.kg-legal.eu/venture-capital-and-private-equity/">Venture Capital &amp; Private Equity</a> <a href="https://www.kg-legal.eu/stock-exchange-and-securities-law-in-poland/">Stock Exchange &amp; Securities</a> <a href="https://www.kg-legal.eu/markets-in-modern-and-conventional-financial-instruments-including-tokens-crypto-assets-packaged-retail-and-insurance-based-investment-products/">Financial Instruments &amp; Crypto</a></td><td><a href="https://www.kg-legal.eu/artificial-intelligence-ai-law-poland-eu/">AI Sector</a> <a href="https://www.kg-legal.eu/data-centers/">Data Centers</a></td><td><a href="https://www.kg-legal.eu/regulatory-law-of-the-energy-sector-in-poland/">Energy Sector</a> <a href="https://www.kg-legal.eu/green-energy-law-renewable-energy-law/">Renewable Energy</a></td><td><a href="https://www.kg-legal.eu/medical-devices-including-medical-devices-law-under-combined-atmps/">Medical Devices</a> <a href="https://www.kg-legal.eu/food-law/">Food Law</a></td><td><a href="https://www.kg-legal.eu/international-divorces/">International Divorces</a> <a href="https://www.kg-legal.eu/international-family-law/">International Family Law</a></td></tr><tr><td><a href="https://www.kg-legal.eu/agreements-and-contracts/">Agreements &amp; Contracts</a> <a href="https://www.kg-legal.eu/contracts-in-the-international-trade-smart-contracts/">International Trade &amp; Smart Contracts</a></td><td><a href="https://www.kg-legal.eu/data-privacy-and-data-protection/">Data Privacy &amp; Protection</a> <a href="https://www.kg-legal.pl/cybersecurity/">Cybersecurity</a></td><td><a href="https://www.kg-legal.eu/biofuels-and-biocomponents/">Biofuels &amp; Biocomponents</a> <a href="https://www.kg-legal.eu/offshore-wind-farms/">Offshore Wind</a></td><td><a href="https://www.kg-legal.eu/art-market/">Art Market</a> <a href="https://www.kg-legal.eu/legal-services-in-fashion-law/">Fashion</a></td><td><a href="https://www.kg-legal.eu/cross-border-inheritance-matters/">Cross Border Inheritance</a> <a href="https://www.kg-legal.eu/immigration-law-business-immigration-polish-passport/">Immigration &amp; Polish Passports</a></td></tr><tr><td><a href="https://www.kg-legal.eu/offshore-companies-tax-optimization/">Offshore Companies &amp; Tax</a></td><td><a href="https://www.kg-legal.eu/distributed-ledger-technology-law/">Distributed Ledger Technology</a> <a href="https://www.kg-legal.eu/metaverse-and-web3-legal-environment/">Metaverse &amp; Web3</a></td><td><a href="https://www.kg-legal.eu/greenfield-in-poland/">Greenfield Investment</a> <a href="https://www.kg-legal.eu/kieltyka-gladkowski-provides-assistance-in-the-area-of-european-partnerships-in-horizon-europe-programme/">Horizon Europe</a></td><td><a href="https://www.kg-legal.pl/cross-border-cases-1/">Cross Border Practice</a> <a href="https://www.kg-legal.eu/customs-and-tariffs-support-for-international-traders-in-poland-and-eu/">Customs &amp; Tariffs</a></td><td><a href="https://www.kg-legal.eu/asset-tracing-in-poland-for-international-creditors/">Asset Tracing in Poland</a> <a href="https://www.kg-legal.eu/asset-tracing/">Asset Tracing</a> <a href="https://www.kg-legal.eu/business-investigations-private-detective-services/">Business Investigations</a></td></tr><tr><td><a href="https://www.kg-legal.eu/dissolution-and-liquidation-of-capital-companies/">Company Dissolution &amp; Liquidation</a> <a href="https://www.kg-legal.eu/cross-border-bankruptcy-in-poland/">Cross Border Bankruptcy</a></td><td><a href="https://www.kg-legal.eu/dual-use-technology-law/">Dual Use Technology</a> <a href="https://www.kg-legal.eu/drone-uas-law/">Drones &amp; UAS</a> <a href="https://www.kg-legal.eu/defence-and-space-sector/">Defence &amp; Space</a></td><td><a href="https://www.kg-legal.eu/industrial-micro-and-public-macro-infrastructure/">Infrastructure</a> <a href="https://www.kg-legal.eu/construction/">Construction</a> <a href="https://www.kg-legal.eu/public-private-partnerships/">Public Private Partnerships</a></td><td><a href="https://www.kg-legal.eu/sworn-certified-legal-translations-of-documents-and-contracts/">Sworn &amp; Certified Legal Translations</a></td><td><a href="https://www.kg-legal.eu/litigation-and-representation-of-international-clients-before-the-polish-courts/">Litigation in Polish Courts</a> <a href="https://www.kg-legal.eu/litigation-antitrust-and-competition-law/">Litigation &amp; Competition</a></td></tr></tbody></table></figure>



<p>Service directory: <a href="https://www.kg-legal.eu/services/">kg-legal.eu/services</a></p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/investment-law-and-processes-in-poland/thuringia-meets-malopolska-kg-legal-kieltyka-gladkowski-at-the-business-meeting-with-minister-president-mario-voigt-at-solaris/">Thuringia Meets Małopolska: KG Legal KIELTYKA GLADKOWSKI at the Business Meeting with Minister-President Mario Voigt at SOLARIS</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/investment-law-and-processes-in-poland/thuringia-meets-malopolska-kg-legal-kieltyka-gladkowski-at-the-business-meeting-with-minister-president-mario-voigt-at-solaris/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Synthetic data generation: the technology and its legal framework in the EU and Poland</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/synthetic-data-generation-the-technology-and-its-legal-framework-in-the-eu-and-poland/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/synthetic-data-generation-the-technology-and-its-legal-framework-in-the-eu-and-poland/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Mon, 16 Nov 2026 16:49:13 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[AI contracts]]></category>
		<category><![CDATA[AI data governance]]></category>
		<category><![CDATA[AI regulation in Poland]]></category>
		<category><![CDATA[AI training data]]></category>
		<category><![CDATA[AI Transparency]]></category>
		<category><![CDATA[AI vendor due diligence]]></category>
		<category><![CDATA[AI-generated content labelling]]></category>
		<category><![CDATA[copyright and AI]]></category>
		<category><![CDATA[data anonymisation]]></category>
		<category><![CDATA[data licensing]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[data protection impact assessment]]></category>
		<category><![CDATA[data provenance]]></category>
		<category><![CDATA[database rights]]></category>
		<category><![CDATA[EU AI Act]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[Generative AI]]></category>
		<category><![CDATA[health data]]></category>
		<category><![CDATA[High-Risk AI Systems]]></category>
		<category><![CDATA[KG Legal]]></category>
		<category><![CDATA[Life Sciences]]></category>
		<category><![CDATA[Polish technology law]]></category>
		<category><![CDATA[pseudonymisation]]></category>
		<category><![CDATA[re-identification risk]]></category>
		<category><![CDATA[SYNTHETIC DATA]]></category>
		<category><![CDATA[synthetic data GDPR compliance]]></category>
		<category><![CDATA[synthetic data in Poland]]></category>
		<category><![CDATA[synthetic data legal framework]]></category>
		<category><![CDATA[synthetic data regulation]]></category>
		<category><![CDATA[synthetic datasets]]></category>
		<category><![CDATA[synthetic images]]></category>
		<category><![CDATA[synthetic tabular data]]></category>
		<category><![CDATA[synthetic text]]></category>
		<category><![CDATA[text and data mining]]></category>
		<category><![CDATA[trade secrets]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=9031</guid>

					<description><![CDATA[<p>Publication date: November 16, 2026 Legal status as at 11 November 2026 Synthetic data, meaning text, images and tabular records produced by algorithms rather than collected from the real world, has moved from the research lab into everyday business use. It promises datasets that can be shared, tested and used to train AI without exposing [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/synthetic-data-generation-the-technology-and-its-legal-framework-in-the-eu-and-poland/">Synthetic data generation: the technology and its legal framework in the EU and Poland</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: November 16, 2026</mark></strong></p>



<p><em>Legal status as at 11 November 2026</em></p>



<p><strong>Synthetic data, meaning text, images and tabular records produced by algorithms rather than collected from the real world, has moved from the research lab into everyday business use. It promises datasets that can be shared, tested and used to train AI without exposing real individuals. The law, however, does not accept that promise at face value. In 2026 the European Data Protection Board stated that synthetic data is not anonymous by default, the AI Act&#8217;s transparency duties for synthetic content began to apply, and Poland established its own AI supervisory authority. This article explains how synthetic data is generated and what legal conditions its generation and use must meet.</strong></p>



<span id="more-9031"></span>



<h1 class="wp-block-heading">What synthetic data is and why the method matters legally</h1>



<p>The term covers three quite different families of techniques, and the legal analysis depends on which one is used.</p>



<ul class="wp-block-list">
<li>Rule-based (mock) data. Values are drawn from dictionaries, formats and random generators, without any real dataset as input. Libraries such as Faker and Mimesis, or older utilities such as pydbgen, produce names, addresses, dates or registration numbers for software testing.</li>



<li>Simulation. Data are produced by a model of the underlying process, for example Monte Carlo sampling or agent-based modelling (the Mesa framework), where simulated agents interact according to defined rules.</li>



<li>Data-derived synthesis. A generative model is trained on a real dataset and then sampled to produce new records that preserve the statistical properties of the original. This is the family that covers large language models, image generators and tabular synthesisers.</li>
</ul>



<p>The first two families usually raise few data protection questions, because no real person&#8217;s data enters the process. The third does: a model that has learned from personal data, copyright works or protected databases carries legal baggage from its training set, and the output may carry it too.</p>



<h1 class="wp-block-heading">Generating text: from Markov chains to large language models</h1>



<p>Natural language processing (NLP) has two branches: natural language understanding (NLU), which extracts meaning from text or speech, and natural language generation (NLG), which produces coherent text from data or a prompt. Both must respect the rules of syntax, vocabulary, morphology and semantics, and generated text should select the relevant information and present it in a logical structure. Typical uses include chatbot and voice assistant responses, suggested replies in e-mail, product descriptions for e-commerce, machine translation, summaries and narrative reports built from financial, business or Internet of Things data, transcription, and paraphrasing.</p>



<p>The methods have developed in four steps.</p>



<ul class="wp-block-list">
<li>Templates and Markov chains. The earliest systems filled the gaps in predefined text templates with data. A Markov chain is a stochastic model in which the probability of the next state depends only on the current one; applied to text, it predicts the next word from the last word entered, which is how early smartphone keyboards suggested words.</li>



<li>Recurrent neural networks (RNNs). An RNN processes a sequence element by element and feeds each output back in as input, so that earlier words are remembered when predicting the next. As sequences grow longer, training breaks down because of vanishing and exploding gradients: the weight updates propagated back through many time steps shrink towards zero, and the network can no longer learn long-range dependencies or produce long, coherent passages.</li>



<li>Long short-term memory (LSTM) networks. A variant of the RNN with three gates (input, forget and output) that decide which information to keep, discard and pass on. Selective memory reduces the vanishing gradient problem, but the architecture is computationally demanding and in practice retains context over a few hundred words.</li>



<li>Transformers. Introduced in 2017, the Transformer combines an encoder-decoder structure with an attention mechanism that relates every word in a sequence to every other word, regardless of distance. Each word keeps its own vector representation instead of the whole sentence being compressed into a single fixed-length vector, and all words are processed in parallel. This made it possible to model long-range dependencies efficiently and is the basis of all leading language models.</li>
</ul>



<p>The best-known Transformer models illustrate the main design choices.</p>



<ul class="wp-block-list">
<li>GPT (generative pre-trained transformer), developed by OpenAI, uses the decoder blocks of the Transformer and predicts the next word from all the preceding ones; the extended sequence is fed back in until the text is complete. GPT-1 (2018) showed that a model pre-trained on unlabelled text could be fine-tuned to specific tasks through transfer learning. GPT-2 and GPT-3 scaled up the data and the number of parameters; GPT-3 was trained on, among other sources, the Common Crawl web corpus.</li>



<li>BERT (bidirectional encoder representations from transformers), developed by Google, uses the encoder. It was trained on BooksCorpus and English Wikipedia with two unsupervised objectives: masked language modelling (MLM), in which randomly hidden words are predicted from the context on both sides, and next sentence prediction (NSP). Lighter or optimised variants include DistilBERT, ALBERT and RoBERTa, and BERT underlies many adaptations to Polish.</li>



<li>XLNet, from Carnegie Mellon University and Google Brain, combines BERT&#8217;s bidirectional context with autoregressive training. Through permutation language modelling it learns from all possible orderings of the words in a sentence, avoiding the artificial masking of inputs, and builds on the Transformer-XL architecture.</li>



<li>T5 (text-to-text transfer transformer), from Google, recasts every NLP task as converting one text string into another, with a prefix indicating the task. It was released in sizes from 60 million to 11 billion parameters and trained on the C4 corpus, a cleaned version of Common Crawl of about 750 GB.</li>



<li>BART, from Facebook AI (now Meta AI), is a denoising autoencoder with a bidirectional encoder and an autoregressive decoder. It learns to reconstruct text corrupted by token masking, token deletion, text infilling, sentence permutation or document rotation, and performs well in summarisation, translation and question answering.</li>
</ul>



<p>Since GPT-3, the GPT-4 generation and its successors, together with open-weight model families, have made fluent, multilingual and multimodal generation a commodity. The remaining weaknesses are no longer stylistic but factual: models can produce plausible but false statements and can reproduce fragments of their training data. For synthetic data purposes, language models are used to generate training dialogues, test documents and paraphrases, and to augment scarce datasets, for example in less widely used languages, where progress has historically lagged behind English.</p>



<h1 class="wp-block-heading">Generating images: autoencoders, GANs, Transformers and diffusion models</h1>



<p>Image generation has become accessible to anyone with a phone and an internet connection. Text-to-image generation combines language understanding with image synthesis: the model must interpret the prompt, map its terms to objects and render shapes, colours and overlapping elements consistently. Synthetic images are widely used to supplement, diversify and balance training sets, including in medical imaging, and in super-resolution, where the resolution of an image is increased while its detail is preserved.</p>



<ul class="wp-block-list">
<li>Variational autoencoders (VAEs). An encoder compresses input images into a latent representation and a decoder generates new images from it. VAEs train quickly and are useful for reducing detail or detecting anomalies, but they render fine detail poorly; today they mostly serve as components of larger systems.</li>



<li>Generative adversarial networks (GANs). A generator produces images from random noise and a discriminator learns to tell them apart from real ones; the two networks improve by competing in a zero-sum game until the discriminator can only guess. GANs are powerful but hard to train and prone to mode collapse, where the generator produces only a narrow range of outputs. Conditional GANs (cGANs) add auxiliary information, such as a class label, to steer generation.</li>



<li>CycleGAN translates images from one domain to another without paired training examples, by learning two mappings that are inverses of each other. Uses include style transfer, changing the season in a photograph and turning paintings into photographs.</li>



<li>StyleGAN, published by Nvidia in 2018, uses a mapping network and a synthesis network to control the style of the generated image at several levels of detail, and is known for highly realistic faces.</li>



<li>Transformer-based models. Image GPT (iGPT) from OpenAI treats an image as a sequence of pixels and completes it in the same way a language model completes a sentence, with impressive results at a high computational cost. The first DALL-E, a 12-billion-parameter model trained on 250 million image-text pairs, combined a Transformer decoder with elements of a variational autoencoder to generate images from text descriptions.</li>



<li>Diffusion models learn to reverse a gradual noising process and generate an image from pure noise by step-by-step denoising, guided by a classifier or, in classifier-free models such as GLIDE, directly by the text. DALL-E 2 (2022) uses the CLIP text and image encoders to produce an image representation that a diffusion decoder turns into a picture. Google&#8217;s Imagen relies on a frozen T5-XXL text encoder and a cascade of diffusion models that increase resolution.</li>
</ul>



<p>DALL-E 2 and Imagen were initially withheld from the public over misuse concerns, but were later released commercially, and openly available models such as Stable Diffusion put the technology in anyone&#8217;s hands. The same approach now extends to video and audio.</p>



<h1 class="wp-block-heading">Generating tabular data and time series</h1>



<p>For structured data, a model of the real dataset is built and then used to draw new values with similar statistical properties, so that an analyst working on the synthetic set reaches the same conclusions as on the original. The methods fall into statistical techniques and machine learning.</p>



<ul class="wp-block-list">
<li>Monte Carlo methods use repeated random sampling from specified probability distributions to model a probabilistic system; by the law of large numbers, the average result converges on the expected value as the number of trials grows. The approach is simple but only as accurate as the empirical data used to define the problem.</li>



<li>Markov chain Monte Carlo (MCMC) and Gibbs sampling generate sequences of dependent observations, which makes it possible to sample from high-dimensional distributions. Gibbs sampling draws each variable in turn from its conditional distribution, which is useful when the joint distribution is unknown or hard to sample directly.</li>



<li>Bayesian networks represent the joint distribution of a set of variables as a directed acyclic graph, with nodes for variables and edges for conditional dependencies. They are computationally efficient and remain a popular way of synthesising population data.</li>



<li>table-GAN adds a third network, a classifier, to the generator and discriminator in order to preserve the semantic integrity of generated records, and was designed with data protection and leak prevention in mind.</li>



<li>CTGAN (conditional tabular GAN) deals with mixed data types, non-Gaussian and multimodal distributions and imbalanced categories through mode-specific normalisation, a conditional generator and training by sampling.</li>



<li>CTAB-GAN adds a classifier and an improved encoding of mixed and highly skewed variables; its successor CTAB-GAN+ incorporates differential privacy, so that utility can be traded against a defined privacy budget.</li>



<li>TimeGAN preserves temporal dynamics in time series such as share prices, using four jointly trained components: an embedding function, a recovery function, a sequence generator and a sequence discriminator.</li>
</ul>



<p>More recent work applies diffusion models and language models to tables. Evaluating the quality of synthetic tabular data remains difficult, because fidelity, usefulness for machine learning and privacy have to be measured separately.</p>



<h1 class="wp-block-heading">Selected tools</h1>



<ul class="wp-block-list">
<li>Faker and Mimesis are Python libraries that generate mock data (names, addresses, dates, payment details and similar fields) for many locales; Mimesis is the faster and more comprehensive of the two. The older pydbgen offers similar functions with output to a Pandas DataFrame, an SQLite table or an Excel file. None of them models relationships found in a real dataset.</li>



<li>Mesa is a framework for agent-based modelling, in which simulated agents (cells, animals, people or organisations) interact with each other and their environment according to defined rules; its DataCollector component records model-level and agent-level variables as tables.</li>



<li>DataSynthesizer takes a real dataset and produces a structurally and statistically similar one with privacy guarantees. Its DataDescriber module learns the distributions and correlations, the DataGenerator samples from them in random, independent-attribute or correlated-attribute mode, and the ModelInspector compares input and output.</li>



<li>Synthetic Data Vault (SDV) is a family of libraries for single-table, multi-table and sequential data, using statistical and deep learning models such as CTGAN, together with metrics for assessing the quality of the result.</li>
</ul>



<p>Before commercial use, the licence terms of each tool and model should be checked, as some are distributed under source-available rather than fully open licences. Every data-derived method also faces the same trade-off: the more faithfully the synthetic set reproduces the original, the more useful it is, and the greater the risk that it reveals something about real individuals. That trade-off is where the legal analysis begins.</p>



<h1 class="wp-block-heading">Is synthetic data personal data?</h1>



<p>The GDPR does not apply to anonymous information (Recital 26). Whether a synthetic dataset qualifies depends on whether individuals in the source data remain identifiable by means reasonably likely to be used. In EDPS v SRB (C-413/23 P, judgment of 4 September 2025) the Court of Justice confirmed that identifiability is assessed in relation to the entity concerned: the same data may be personal for the party holding the key or the original records and not personal for a recipient with no realistic means of identification.</p>



<p>On 7 July 2026 the European Data Protection Board adopted draft Guidelines 02/2026 on anonymisation, the first EU-level update since the Article 29 Working Party&#8217;s Opinion 05/2014. The public consultation closed on 30 October 2026 and the final text is awaited. The draft matters for synthetic data in four respects.</p>



<ul class="wp-block-list">
<li>No presumption of anonymity. AI model outputs and synthetic data are not anonymous by default. The risk of re-identification need not be zero, but it must be insignificant in reality.</li>



<li>Three criteria. Anonymity is tested against record isolation, linkage and inference. For synthetic data the inference test is the critical one: if querying the data or the generator reveals something specific about a person, or reveals that the person was in the source dataset, anonymity is in doubt. General knowledge about a population does not count.</li>



<li>Generation is processing. Training a generator on personal data is itself processing and needs a legal basis under Article 6 GDPR and, for health or other special category data, an exemption under Article 9(2), even if the output is anonymous. Data subjects should be informed that their data will be used in this way.</li>



<li>Documentation and review. The controller must be able to demonstrate anonymity, keep a record of the tests performed and reassess when technology lowers the cost of re-identification. A contractual ban on re-identification helps but is not sufficient on its own.</li>
</ul>



<p>The same reasoning applies to the generative model. In Opinion 28/2024 the EDPB concluded that an AI model trained on personal data cannot automatically be regarded as anonymous, because personal data may be extracted from it. Companies that buy or license a synthesiser trained on someone else&#8217;s data should therefore ask how it was trained and tested.</p>



<p>The legislative picture is still moving. The Commission&#8217;s Digital Omnibus proposal of November 2025 would have narrowed the definition of personal data, but the Council&#8217;s negotiating mandate of 21 October 2026 keeps the current definition and instead proposes a specific rule on pseudonymised data and a clearer path for relying on legitimate interest in AI development. The European Parliament has not yet adopted its position, so none of this is law.</p>



<h1 class="wp-block-heading">Synthetic data under the AI Act</h1>



<p>Regulation (EU) 2024/1689 (the AI Act) treats synthetic data as a privacy-preserving alternative, not as an exemption. Three points are worth noting.</p>



<ul class="wp-block-list">
<li>Data governance. Training, validation and testing datasets for high-risk AI systems must be relevant, sufficiently representative and, as far as possible, free of errors, and their origin must be documented (Article 10). These requirements apply equally to synthetic datasets, which can inherit or amplify the biases of their source.</li>



<li>Synthetic data first. Under the new Article 4a, introduced by Regulation (EU) 2026/1744 in place of the former Article 10(5), special categories of personal data may be processed to detect and correct bias only where this cannot be achieved with other data, including synthetic or anonymised data. Article 59 applies the same logic to further processing of personal data in AI regulatory sandboxes.</li>



<li>Revised timetable. The same amending regulation, in force since 27 July 2026, postponed the high-risk obligations to 2 December 2027 for Annex III systems and 2 August 2028 for systems covered by Annex I product legislation. Obligations for providers of general-purpose AI models have applied since 2 August 2025.</li>
</ul>



<h1 class="wp-block-heading">Synthetic content: marking and labelling duties</h1>



<p>The AI Act also regulates the output side. Since 2 August 2026, under Article 50(2), providers of AI systems that generate synthetic audio, image, video or text must ensure that outputs are marked in a machine-readable format and are detectable as artificially generated. Systems already on the market before that date have until 2 December 2026 to comply. Under Article 50(4), deployers must disclose deep fakes and AI-generated text published to inform the public on matters of public interest, unless the text has undergone human review and is under editorial responsibility.</p>



<p>The Commission published a voluntary Code of Practice on marking and labelling of AI-generated content on 10 June 2026 and final guidelines on Article 50 on 20 July 2026. According to the guidelines, marking is not required, among other cases, for outputs that stay within an organisation in controlled business settings, a point of practical importance for synthetic datasets used purely internally for testing or model training. In addition, from 2 December 2026 the AI Act prohibits AI systems that generate non-consensual intimate imagery of identifiable persons.</p>



<h1 class="wp-block-heading">Copyright and database rights</h1>



<p>A generator trained on protected works or databases raises intellectual property questions at two stages.</p>



<p>Training. Reproductions made for text and data mining are permitted under Articles 3 and 4 of Directive (EU) 2019/790, the latter only where the rightholder has not reserved its rights in an appropriate manner. Poland transposed these exceptions in Articles 26² and 26³ of the Copyright and Related Rights Act with effect from 20 September 2024. Providers of general-purpose AI models must in addition maintain a copyright compliance policy and publish a summary of training content (Article 53 of the AI Act). The first case on generative AI before the Court of Justice, Like Company v Google (C-250/25), is pending. In his Opinion of 3 November 2026 Advocate General Szpunar advised the Court to declare the questions on model training inadmissible, while taking the view that real-time retrieval and reuse of press articles in chatbot answers may infringe press publishers&#8217; rights. The judgment is still to come, and first-instance rulings in national courts, such as Getty Images v Stability AI in London and GEMA v OpenAI in Munich, point in different directions.</p>



<p>Output. Purely machine-generated content is generally not protected by copyright in the EU or in Poland, because protection requires a human author&#8217;s creative contribution. A synthetic dataset may therefore need to be protected by contract and as a trade secret. Conversely, output that reproduces recognisable parts of a training work may infringe, and extraction of a substantial part of a protected database may engage the sui generis right under Directive 96/9/EC.</p>



<h1 class="wp-block-heading">Health data and other regulated sectors</h1>



<p>Synthetic data is particularly attractive in life sciences, where patient-level data is scarce and sensitive. Health data used to train a generator is special category data, so an Article 9(2) GDPR ground, a data protection impact assessment and robust re-identification testing will normally be required. The European Health Data Space Regulation (EU) 2025/327, whose rules on secondary use start to apply from March 2029, will create a formal route for access to health data for research and innovation. Where synthetic data supports the development or validation of a medical device or a regulated financial model, sectoral regulators will also expect evidence that it is representative of the real population.</p>



<h1 class="wp-block-heading">The Polish dimension</h1>



<p>The Act of 3 July 2026 on Artificial Intelligence Systems, which for the most part entered into force on 11 August 2026, established the Commission for the Development and Safety of Artificial Intelligence (KRiBSI) as the market surveillance authority and single point of contact under the AI Act. KRiBSI can conduct inspections, hear complaints, impose administrative fines and issue individual opinions on the application of the AI Act, and the Act provides a basis for regulatory sandboxes. The President of the Personal Data Protection Office (UODO) remains competent for the data protection aspects, so a project involving data-derived synthetic data may fall under the supervision of both authorities.</p>



<h1 class="wp-block-heading">Practical checklist</h1>



<ul class="wp-block-list">
<li>Identify the method: rule-based, simulated or derived from real data. Only the last normally triggers the GDPR analysis.</li>



<li>Establish the legal basis and purpose compatibility for using the source data to train the generator, and inform data subjects.</li>



<li>Test the output and the model against record isolation, linkage and inference, document the results and schedule reassessment.</li>



<li>Do not label a dataset as anonymous in contracts or privacy notices unless the assessment supports it.</li>



<li>Check the provenance and licence of the generator, the training data and any pre-trained model, including text and data mining reservations.</li>



<li>Where synthetic data feeds a high-risk AI system, document its origin, representativeness and known limitations in line with Article 10 of the AI Act.</li>



<li>Where synthetic content is released outside the organisation, apply the marking and labelling rules of Article 50.</li>



<li>Allocate responsibility in contracts: warranties on training data, re-identification prohibitions, audit rights and liability for infringing output.</li>
</ul>



<h1 class="wp-block-heading">Key sources</h1>



<ol class="wp-block-list">
<li>Regulation (EU) 2016/679 (GDPR), in particular Articles 4(1), 6, 9 and 35 and Recital 26.</li>



<li>Judgment of the Court of Justice of 4 September 2025, EDPS v SRB, C-413/23 P.</li>



<li>EDPB, draft Guidelines 02/2026 on anonymisation (adopted for consultation on 7 July 2026); EDPB Opinion 28/2024 on AI models.</li>



<li>Regulation (EU) 2024/1689 (AI Act), in particular Articles 4a, 10, 50, 53 and 59, as amended by Regulation (EU) 2026/1744.</li>



<li>Code of Practice on marking and labelling of AI-generated content (10 June 2026) and Commission Guidelines on Article 50 of the AI Act (20 July 2026).</li>



<li>Directive (EU) 2019/790, Articles 3 and 4; Directive 96/9/EC; Polish Act of 4 February 1994 on Copyright and Related Rights.</li>



<li>Opinion of Advocate General Szpunar of 3 November 2026, Like Company v Google Ireland, C-250/25.</li>



<li>Regulation (EU) 2025/327 on the European Health Data Space.</li>



<li>Polish Act of 3 July 2026 on Artificial Intelligence Systems.</li>
</ol>



<p><em>This text is for information purposes only and does not constitute legal advice.</em></p>
<p> </p>



<p>Services</p>



<p><strong>KIELTYKA GLADKOWSKI</strong> is a Polish business law firm headquartered in Warsaw and Krakow, with a New York office under the iSTART1 brand. Building on a practice dating back to 1981, we advise clients in the following areas.</p>



<h1 class="wp-block-heading">Explore our services</h1>



<p>Select a practice area to learn more.</p>



<figure class="wp-block-table"><table class="has-luminous-vivid-amber-background-color has-background has-fixed-layout"><thead><tr><td><strong>Business<br>&amp; Finance</strong></td><td><strong>Technology<br>&amp; IP</strong></td><td><strong>Energy<br>&amp; Projects</strong></td><td><strong>Sectors &amp;<br>Cross Border</strong></td><td><strong>Private Clients<br>&amp; Disputes</strong></td></tr></thead><tbody><tr><td><a href="https://www.kg-legal.pl/investment-law-and-processes-in-poland-1/">Commercial &amp; Corporate</a> <a href="https://www.kg-legal.eu/polish-limited-liability-company-legal-support-in-operational-business-maintenance-and-representation-of-shareholders-in-court/">Polish LLCs</a> <a href="https://www.kg-legal.eu/employment-law-and-collective-labour-law-in-poland-including-employment-contracts/">Employment Law</a></td><td><a href="https://www.kg-legal.pl/it-bio-it-new-technologies-media-law-1/">IT, IP, Media &amp; New Tech</a> <a href="https://www.kg-legal.eu/trademarks-and-brand-protection/">Trademarks &amp; Brand Protection</a></td><td><a href="https://www.kg-legal.eu/green-transition/">Green Transition</a> <a href="https://www.kg-legal.eu/energy-transformation-in-poland-main-assumptions-of-the-polish-energy-policy-2040-2/">Energy Policy 2040</a></td><td><a href="https://www.kg-legal.pl/pharmaceutical-life-sciences-healthcare-law-1/">Pharma, Life Sciences &amp; Healthcare</a> <a href="https://www.kg-legal.eu/pharmaceutical-law/">Pharmaceutical Law</a></td><td><a href="https://www.kg-legal.eu/private-client-desk/">Private Client Desk</a> <a href="https://www.kg-legal.eu/hnw-and-next-gen-wealth/">HNW &amp; Next Gen Wealth</a></td></tr><tr><td><a href="https://www.kg-legal.eu/venture-capital-and-private-equity/">Venture Capital &amp; Private Equity</a> <a href="https://www.kg-legal.eu/stock-exchange-and-securities-law-in-poland/">Stock Exchange &amp; Securities</a> <a href="https://www.kg-legal.eu/markets-in-modern-and-conventional-financial-instruments-including-tokens-crypto-assets-packaged-retail-and-insurance-based-investment-products/">Financial Instruments &amp; Crypto</a></td><td><a href="https://www.kg-legal.eu/artificial-intelligence-ai-law-poland-eu/">AI Sector</a> <a href="https://www.kg-legal.eu/data-centers/">Data Centers</a></td><td><a href="https://www.kg-legal.eu/regulatory-law-of-the-energy-sector-in-poland/">Energy Sector</a> <a href="https://www.kg-legal.eu/green-energy-law-renewable-energy-law/">Renewable Energy</a></td><td><a href="https://www.kg-legal.eu/medical-devices-including-medical-devices-law-under-combined-atmps/">Medical Devices</a> <a href="https://www.kg-legal.eu/food-law/">Food Law</a></td><td><a href="https://www.kg-legal.eu/international-divorces/">International Divorces</a> <a href="https://www.kg-legal.eu/international-family-law/">International Family Law</a></td></tr><tr><td><a href="https://www.kg-legal.eu/agreements-and-contracts/">Agreements &amp; Contracts</a> <a href="https://www.kg-legal.eu/contracts-in-the-international-trade-smart-contracts/">International Trade &amp; Smart Contracts</a></td><td><a href="https://www.kg-legal.eu/data-privacy-and-data-protection/">Data Privacy &amp; Protection</a> <a href="https://www.kg-legal.pl/cybersecurity/">Cybersecurity</a></td><td><a href="https://www.kg-legal.eu/biofuels-and-biocomponents/">Biofuels &amp; Biocomponents</a> <a href="https://www.kg-legal.eu/offshore-wind-farms/">Offshore Wind</a></td><td><a href="https://www.kg-legal.eu/art-market/">Art Market</a> <a href="https://www.kg-legal.eu/legal-services-in-fashion-law/">Fashion</a></td><td><a href="https://www.kg-legal.eu/cross-border-inheritance-matters/">Cross Border Inheritance</a> <a href="https://www.kg-legal.eu/immigration-law-business-immigration-polish-passport/">Immigration &amp; Polish Passports</a></td></tr><tr><td><a href="https://www.kg-legal.eu/offshore-companies-tax-optimization/">Offshore Companies &amp; Tax</a></td><td><a href="https://www.kg-legal.eu/distributed-ledger-technology-law/">Distributed Ledger Technology</a> <a href="https://www.kg-legal.eu/metaverse-and-web3-legal-environment/">Metaverse &amp; Web3</a></td><td><a href="https://www.kg-legal.eu/greenfield-in-poland/">Greenfield Investment</a> <a href="https://www.kg-legal.eu/kieltyka-gladkowski-provides-assistance-in-the-area-of-european-partnerships-in-horizon-europe-programme/">Horizon Europe</a></td><td><a href="https://www.kg-legal.pl/cross-border-cases-1/">Cross Border Practice</a> <a href="https://www.kg-legal.eu/customs-and-tariffs-support-for-international-traders-in-poland-and-eu/">Customs &amp; Tariffs</a></td><td><a href="https://www.kg-legal.eu/asset-tracing-in-poland-for-international-creditors/">Asset Tracing in Poland</a> <a href="https://www.kg-legal.eu/asset-tracing/">Asset Tracing</a> <a href="https://www.kg-legal.eu/business-investigations-private-detective-services/">Business Investigations</a></td></tr><tr><td><a href="https://www.kg-legal.eu/dissolution-and-liquidation-of-capital-companies/">Company Dissolution &amp; Liquidation</a> <a href="https://www.kg-legal.eu/cross-border-bankruptcy-in-poland/">Cross Border Bankruptcy</a></td><td><a href="https://www.kg-legal.eu/dual-use-technology-law/">Dual Use Technology</a> <a href="https://www.kg-legal.eu/drone-uas-law/">Drones &amp; UAS</a> <a href="https://www.kg-legal.eu/defence-and-space-sector/">Defence &amp; Space</a></td><td><a href="https://www.kg-legal.eu/industrial-micro-and-public-macro-infrastructure/">Infrastructure</a> <a href="https://www.kg-legal.eu/construction/">Construction</a> <a href="https://www.kg-legal.eu/public-private-partnerships/">Public Private Partnerships</a></td><td><a href="https://www.kg-legal.eu/sworn-certified-legal-translations-of-documents-and-contracts/">Sworn &amp; Certified Legal Translations</a></td><td><a href="https://www.kg-legal.eu/litigation-and-representation-of-international-clients-before-the-polish-courts/">Litigation in Polish Courts</a> <a href="https://www.kg-legal.eu/litigation-antitrust-and-competition-law/">Litigation &amp; Competition</a></td></tr></tbody></table></figure>



<p>Service directory: <a href="https://www.kg-legal.eu/services/">kg-legal.eu/services</a></p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/synthetic-data-generation-the-technology-and-its-legal-framework-in-the-eu-and-poland/">Synthetic data generation: the technology and its legal framework in the EU and Poland</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/synthetic-data-generation-the-technology-and-its-legal-framework-in-the-eu-and-poland/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Ageing biomarkers as evidence of geroprotective efficacy</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ageing-biomarkers-as-evidence-of-geroprotective-efficacy/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ageing-biomarkers-as-evidence-of-geroprotective-efficacy/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Tue, 10 Nov 2026 16:55:21 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[Ageing Biomarkers]]></category>
		<category><![CDATA[Biobanking]]></category>
		<category><![CDATA[Biological Age]]></category>
		<category><![CDATA[Biological Samples]]></category>
		<category><![CDATA[Biomarker Validation]]></category>
		<category><![CDATA[Clinical Evidence]]></category>
		<category><![CDATA[Clinical Trial Design]]></category>
		<category><![CDATA[clinical trials]]></category>
		<category><![CDATA[Clinical Trials Regulation]]></category>
		<category><![CDATA[Companion Diagnostics]]></category>
		<category><![CDATA[CTIS]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[EMA]]></category>
		<category><![CDATA[Epigenetic Clocks]]></category>
		<category><![CDATA[EU Law]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[Genetic Data]]></category>
		<category><![CDATA[Geroprotection]]></category>
		<category><![CDATA[Geroprotectors]]></category>
		<category><![CDATA[Health Claims]]></category>
		<category><![CDATA[Healthy Ageing]]></category>
		<category><![CDATA[In Vitro Diagnostics]]></category>
		<category><![CDATA[informed consent]]></category>
		<category><![CDATA[Intellectual Property]]></category>
		<category><![CDATA[IVDR]]></category>
		<category><![CDATA[Life Sciences Law]]></category>
		<category><![CDATA[Longevity Research]]></category>
		<category><![CDATA[Medical Device Software]]></category>
		<category><![CDATA[Non-Commercial Clinical Trials]]></category>
		<category><![CDATA[Patent Protection]]></category>
		<category><![CDATA[Performance Studies]]></category>
		<category><![CDATA[Polish law]]></category>
		<category><![CDATA[Research Commercialisation]]></category>
		<category><![CDATA[Research Ethics]]></category>
		<category><![CDATA[Surrogate Endpoints]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8985</guid>

					<description><![CDATA[<p>Publication date: November 10, 2026 Clinical research, diagnostic testing and commercialisation under Polish and EU law Law as at 5 November 2026 Abstract A change in an epigenetic clock may justify further investigation of a medicine. Its value as evidence of benefit depends on what the clock measures, the population in which it has been [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ageing-biomarkers-as-evidence-of-geroprotective-efficacy/">Ageing biomarkers as evidence of geroprotective efficacy</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color"><strong>Publication date: November 10, 2026</strong></mark></p>



<p>Clinical research, diagnostic testing and commercialisation under Polish and EU law</p>



<p><em>Law as at 5 November 2026</em></p>



<p>Abstract</p>



<p>A change in an epigenetic clock may justify further investigation of a medicine. Its value as evidence of benefit depends on what the clock measures, the population in which it has been validated and whether treatment effects on the score predict effects on health. The same assay may be used to analyse archived samples, select participants for treatment or sell a test to consumers. Each use raises a different legal question. This article traces those questions from trial design through to commercial use, with particular attention to Polish law on medical experiments, performance studies, access to medical records and the use of data from non-commercial clinical trials.</p>



<p><strong>Keywords&nbsp; </strong>ageing biomarkers, geroprotectors, clinical trials, IVDR, biobanks, genetic data</p>



<span id="more-8985"></span>



<h2 class="wp-block-heading">What three years on a biological clock mean</h2>



<p>Suppose researchers measure DNA methylation in volunteers’ blood before administering a substance and again six months later. In the intervention group, the mean score on the chosen clock falls by three years. The sponsor wants to describe the result as evidence that the product slows ageing. The first question concerns the comparator. Was the difference from placebo also three years, or does the figure describe only the change between two measurements in the same people? The second concerns the volunteers. Does a change of this size predict a later loss of independence, fewer hospital admissions or longer survival? Each claim calls for different evidence.<a href="#_ftn1" id="_ftnref1">[1]</a></p>



<p>A geroprotector, as used here, is a substance under investigation for its potential effect on biological processes associated with ageing. A model can be trained to estimate chronological age, mortality risk or the pace of physiological change. Two models that express their results in years can therefore be measuring different things. In an open-label pilot study in people with diabetic kidney disease, investigators reported fewer cells expressing markers of cellular senescence, including p16 and p21, in adipose tissue after dasatinib and quercetin. This provided evidence about a possible mechanism of these senolytic drugs. Establishing whether the intervention prolongs healthy life requires a study measuring that benefit or a reliable surrogate for it.<a href="#_ftn2" id="_ftnref2">[2]</a></p>



<p>The European Innovation Council recognised this problem in its Pathfinder challenge on Biotechnology for Healthy Ageing. For biomarker tools, the programme linked access to appropriate cohorts with the need to demonstrate a relationship between an ageing mechanism and clinical features. These conditions identify a research objective. Whether the results support a particular medicinal indication is a separate regulatory assessment.<a href="#_ftn3" id="_ftnref3">[3]</a></p>



<h1 class="wp-block-heading">From a biomarker association to a clinical benefit</h1>



<p>An ageing biomarker may identify a group at increased risk of functional decline. Such a prognostic marker makes events more likely to be observed during the study. A predictive marker helps identify patients likely to derive a greater effect from a particular treatment. A pharmacodynamic response marker serves another purpose. A change may show that the substance affects the biological pathway under investigation. If the sponsor uses that change to predict something that matters to patients, the marker is being proposed as a surrogate endpoint. The strength of that case depends on the biological rationale, the marker’s prognostic value and interventional evidence that treatment effects on the marker correspond to treatment effects on the clinical outcome.<a href="#_ftn4" id="_ftnref4">[4]</a></p>



<p>The CALERIE trial illustrates the distinction. In a post hoc analysis of samples from this randomised study of caloric restriction, the intervention slowed the pace of ageing measured by DunedinPACE. The authors found no significant change in the biological age estimates produced by PhenoAge and GrimAge. In a separate study, TranslAGE researchers applied 16 clocks to data from 51 interventional studies. Responsiveness varied with the intervention, population and length of follow-up. For a lawyer reviewing a regulatory dossier, the choice of clock is therefore material. Selecting the most favourable measure after inspecting the results weakens the inference, particularly when the investigators have made numerous comparisons. Whether the population studied resembles the population for the intended efficacy claim is a further question.<a href="#_ftn5" id="_ftnref5">[5]</a></p>



<p>A protocol should specify the algorithm version, tissue, sampling time, primary comparison and approach to multiplicity and missing measurements. A study of loss of independence also has to address what happens if a participant dies before the next functional assessment. An analysis confined to those who are well enough to attend the visit could give an overly favourable picture. ICH E9(R1) provides a way to link these decisions to the precise treatment effect the trial is intended to estimate. Choices made on this point can change the clinical meaning of the answer even if the assay is perfectly reproducible.<a href="#_ftn6" id="_ftnref6">[6]</a></p>



<p>Where a method is intended for use by several medicine developers, the European Medicines Agency (EMA) offers voluntary qualification within a defined context of use. Its Committee for Medicinal Products for Human Use (CHMP) may, for example, assess whether the tool is suitable for identifying a higher-risk population. Using the same tool as a primary endpoint requires a case for that additional purpose. Where the question concerns the design of one product development programme, the sponsor can seek scientific advice on the protocol.<a href="#_ftn7" id="_ftnref7">[7]</a></p>



<h1 class="wp-block-heading">Classifying the substance and choosing the study design</h1>



<p>The biological nature of ageing does not by itself determine product classification or establish a proposed indication. Under EU and Polish law, a substance may be a medicinal product because it is presented as treating or preventing disease. It may also qualify because it can be used to restore, correct or modify physiological functions through pharmacological, immunological or metabolic action. In Hecht-Pharma, the Court of Justice held that the product must be assessed individually, taking account of factors including composition, dose and effects under normal conditions of use. Giving medicines legislation priority in a borderline case still requires a finding that the product meets the definition of a medicine.<a href="#_ftn8" id="_ftnref8">[8]</a></p>



<p>This distinction affects population selection. A study of whether treatment delays mobility loss in people at increased risk can define an event and a period of observation. A project in healthy adults that measures only a cellular marker will answer a narrower question. Experience with metformin in a different indication contributes knowledge about the substance. A new use still requires its own assessment of dose, duration, population and benefit–risk balance. A clinical trial authorisation allows the evidence to be generated. Bringing a new medicine to market or adding an indication to an existing authorisation requires a separate procedure. The authority considers, among other matters, whether therapeutic efficacy is substantiated and the benefit–risk balance is favourable for the proposed use.<a href="#_ftn9" id="_ftnref9">[9]</a></p>



<p>The interventions in the protocol then determine the legal route. Under Regulation (EU) No 536/2014 (the Clinical Trials Regulation, or CTR), a clinical study of a medicinal product constitutes a clinical trial if any one of the statutory criteria is satisfied. They include advance assignment to a therapeutic strategy outside normal practice, a prescribing decision taken together with the decision to enrol the participant or diagnostic or monitoring procedures additional to normal care. Observing treatment prescribed independently of enrolment can constitute a non-interventional study if the other criteria are also absent. Extra blood draws required by the protocol therefore merit scrutiny even if the medicine is already available from a pharmacy.<a href="#_ftn10" id="_ftnref10">[10]</a></p>



<p>A low-intervention trial is still a clinical trial. Apart from placebos, the investigational medicines must be authorised. Their use must conform to the authorisation or be supported by published scientific evidence of safety and efficacy in at least one Member State concerned. Additional procedures must pose no more than minimal additional risk or burden compared with normal clinical practice. An earlier stage of a programme may involve animal experiments. In Poland, these require authorisation by the competent local ethics committee. The assessment considers replacement of animals by other methods, reduction of animal numbers and refinement of procedures.<a href="#_ftn11" id="_ftnref11">[11]</a></p>



<h1 class="wp-block-heading">Protecting participants in a Polish clinical trial</h1>



<p>Applications for medicinal product trials are submitted through the EU Clinical Trials Information System (CTIS). In a multinational trial, Part I of the assessment covers matters including the scientific rationale and benefit–risk balance. In Part II, each Member State concerned considers national matters including participant information, consent, recruitment and arrangements for compensation. The Polish decision is issued by the President of the Office for Registration of Medicinal Products, Medical Devices and Biocidal Products (URPL). The ethical assessment is carried out by the national bioethics committee or a designated committee on the statutory list.<a href="#_ftn12" id="_ftnref12">[12]</a></p>



<p>A healthy volunteer may have little prospect of a direct benefit from a preventive intervention. The application should explain what potential value to participants or public health justifies the foreseeable risks and inconvenience. In older populations, multimorbidity, concurrent medicines and the burden of study visits affect that assessment. EMA recommends the Short Physical Performance Battery to characterise physical frailty at baseline. It helps establish whom the study population represents. A proposal to use the same scale as the main measure of efficacy needs a justification appropriate to the indication.<a href="#_ftn13" id="_ftnref13">[13]</a></p>



<p>The English MET-PREVENT trial shows why that distinction matters. In older people with probable sarcopenia and physical prefrailty or frailty, metformin did not improve four-metre walking speed at four months compared with placebo. The authors also reported poor tolerability in this population. That finding concerns a particular dose, population and functional outcome. In another trial, a promising change in a laboratory marker would still have to be interpreted alongside tolerability, adverse events and the participant’s ability to function in daily life.<a href="#_ftn14" id="_ftnref14">[14]</a></p>



<p>Chronological age does not determine the ability to give informed consent. For participants who lack that capacity, Article 31 CTR requires, among other safeguards, consent from a legally designated representative, respect for the express wish of a participant capable of forming an opinion and a justification for studying that population. The sponsor must also plan safety oversight, reporting of suspected unexpected serious adverse reactions and publication of results with a lay summary, generally within one year of the end of the trial in all Member States concerned. Polish law addresses civil liability and insurance, with an exception to the sponsor’s insurance obligation for low-intervention trials. The Clinical Trials Compensation Fund provides a separate route to compensation, including for bodily injury or a health impairment resulting from trial participation.<a href="#_ftn15" id="_ftnref15">[15]</a></p>



<h1 class="wp-block-heading">When a biomarker assay falls within the IVDR</h1>



<p>Regulation (EU) 2017/746 (the IVDR) governs devices intended to examine specimens from the human body, including software with the relevant intended purpose. The biological properties of the marker and the intended use of the tool that measures it are distinct issues. A retrospective analysis of blinded samples may examine whether changes in methylation correlate with subsequent functional decline. Where the result plays no part in participant management, the assay may have an exclusively research purpose. If the same result determines enrolment, assignment to a medicine or discontinuation of treatment, a decision about an individual now depends on it. Assessment must take account of the manufacturer’s intended purpose, instructions, promotional statements and actual role in the protocol. A “research use only” label cannot resolve the issue on its own.<a href="#_ftn16" id="_ftnref16">[16]</a></p>



<p>Consider the clock again. Dividing participants into subgroups for later statistical analysis has different implications from withholding the medicine from an individual whose score falls below a threshold. Companion diagnostic (CDx) status has a narrower test. The device must be essential for the safe and effective use of a corresponding medicine and identify people likely to benefit or those at increased risk of serious adverse reactions. As a rule, a companion diagnostic is a class C device under IVDR rule 3(f). In the conformity assessment, the notified body consults a competent medicines authority or EMA, as appropriate for the medicinal product.<a href="#_ftn17" id="_ftnref17">[17]</a></p>



<p>The performance evaluation addresses the scientific validity of the association between the measured feature and a clinical condition, analytical performance and clinical performance. An assay can measure methylation accurately yet assign patients to the wrong treatment group, for example if the threshold was derived in a younger population. Documentation should address the consequences of false positive and false negative results. The intended use covered by an existing CE mark must be checked against the proposed new purpose. Evidence that the assay performs as intended does not establish that the medicine works.<a href="#_ftn18" id="_ftnref18">[18]</a></p>



<p>Under the standard Polish procedure, a medicinal product trial and a device performance study requiring authorisation are assessed separately. URPL specifies an ethics opinion and authorisation by its President for a CDx performance study using samples collected specifically for that purpose. CDx studies using only left-over samples are subject to notification. The EU COMBINE pilot coordinates reviews of selected multinational combined studies. Poland is not listed among the participating countries in Annex I to the phase two sponsor guide. A sponsor therefore needs to plan the Polish national route even where sites elsewhere in the same project participate in the pilot.<a href="#_ftn19" id="_ftnref19">[19]</a></p>



<h1 class="wp-block-heading">Software and the result delivered to a patient</h1>



<p>An ageing clock may be a statistical model, stand-alone medical software or a component of a diagnostic system. A program designed to interpret specimen results and provide medical information about an individual may fall under the IVDR. Software that draws on gait sensors and supports a diagnostic or therapeutic decision may instead fall under Regulation (EU) 2017/745 (the MDR). For a model that combines laboratory results, images and clinical history, the function of each module and the principal source of the information used for the decision matter. The manufacturer’s description should be tested against what users actually receive and what the product invites them to do.<a href="#_ftn20" id="_ftnref20">[20]</a></p>



<p>An application might provide general educational information on ageing. A different question arises if its report states that the customer’s profile warrants a particular therapy. Before deployment, the developer needs to establish the limits of validation, how uncertainty is communicated and how the algorithm performs across age groups and in people with multiple conditions. An update to the model or threshold may change how many patients are offered treatment. Preserving the trial version, recording changes and evaluating their effects makes it possible to reconstruct the basis for an individual decision.</p>



<p>If the software also meets the definition of an AI system, Article 6(1) of the AI Act requires a separate assessment. It covers an AI system that is a product, or a safety component of a product, subject to Union legislation listed in Annex I where third-party conformity assessment is required for that product. Regulation (EU) 2026/1744 moved the application of the high-risk requirements for this group to 2 August 2028. An exclusion for systems developed and put into service solely for scientific research and development requires reassessment when the system becomes part of a clinical offering. Medical device requirements apply on their own timetable.<a href="#_ftn21" id="_ftnref21">[21]</a></p>



<p>Polish laboratory medicine law regulates the activities of medical laboratories and the reporting of results to patients. Relevant duties concern staff qualifications, specimen identification, method validation, authorisation of results and quality standards. Genetic testing in healthcare may also engage Article 4 IVDR, including information about the nature and implications of testing and access to counselling for specified predispositions to conditions generally considered untreatable. If a hospital makes and uses its own assay, Article 5(5) IVDR calls for a separate assessment, including use within the health institution, no transfer of the device to another legal entity and an appropriate quality management system. Ordering a test online does not establish its regulatory status or the qualifications needed to interpret its result.<a href="#_ftn22" id="_ftnref22">[22]</a></p>



<h1 class="wp-block-heading">Human samples and the limits of biobanking</h1>



<p>The Polish Act on the Professions of Physician and Dentist treats research on biological material, including genetic material, taken from a person for scientific purposes as a medical experiment (Article 21(4)). Article 29a(2) deals separately with collection, processing, storage and distribution of material for scientific purposes. These biobanking activities are excluded from the definition. A biobank may thus preserve a sample under its procedures, while a subsequent project testing a specific hypothesis on that sample may require a favourable bioethics opinion, consent and insurance. Under this Polish regime a research experiment in a healthy or ill person may entail at most minimal risk proportionate to its potential scientific benefit. Article 23a excludes, among others, persons subject to a legal incapacitation order from such research experiments. These national conditions are distinct from the rules specific to medicinal product and device studies.<a href="#_ftn23" id="_ftnref23">[23]</a></p>



<p>Consider two plasma specimens stored in the same freezer. The first was drawn to establish an ageing research cohort and can be linked to its donor. The second is an anonymous remnant of routine diagnosis. In its TRANSMED guidance, the Polish Medical Research Agency distinguishes the original purpose of collection, the investigator’s access to the key and subsequent use of data in a clinical trial. These are practical pointers for assessing an individual project. Placing material in a biobank does not confer rights to carry out every future analysis. Article 28 of the Polish Act provides a particular route for scientific use of information from a medical experiment or screening where identification is impossible. That provision addresses information.<a href="#_ftn24" id="_ftnref24">[24]</a></p>



<p>Under Article 24(2)(7), participants in a medical experiment must be informed about foreseeable further use of results, data and material, including commercial use. For a sample bank, the practical decisions are whether new markers can be assayed, whether a specimen can be sent abroad, whether donors can be recontacted and whether an incidental finding with possible health implications will be returned. A research observation may first require clinical confirmation. The participant also needs a meaningful account of what withdrawal entails if material is still held, has been consumed in testing or has already contributed to aggregate results.<a href="#_ftn25" id="_ftnref25">[25]</a></p>



<p>In findings published on 22 October 2026, Poland’s Supreme Audit Office described genetic samples being sent to foreign laboratories without patients’ knowledge or consent. In a multicentre project, a material transfer agreement should identify the specimens and recipients, permitted analyses, conditions for onward transfer, return or destruction, and responsibility for breaching those limits. The parties must also establish the sender’s authority and check that the proposed transfer is consistent with what the donor was told. A signed agreement between institutions cannot supply permissions that neither institution holds.<a href="#_ftn26" id="_ftnref26">[26]</a></p>



<h1 class="wp-block-heading">Participant data and access to European cohorts</h1>



<p>A clock score linked to a participant may reveal information about present or future health. DNA analysis can also entail processing genetic data. The controller must identify each purpose and the relevant lawful basis under Article 6 and condition under Article 9(2) GDPR. Reliance on the scientific research condition in Article 9(2)(j) requires a basis in Union or Member State law and safeguards under Article 89. Coding limits the visibility of a participant’s identity. Where reidentification remains reasonably possible, the data remain protected. This matters when rare variants are combined with age, clinical history or a small recruitment site.<a href="#_ftn27" id="_ftnref27">[27]</a></p>



<p>The European Data Protection Board distinguishes informed consent to trial participation from the legal basis for processing personal data. The sponsor’s duties to ensure safety and reliability serve a different purpose from scientific analysis. Withdrawal ends further interventions covered by the participant’s decision, but does not automatically remove data already lawfully obtained. Article 58 CTR requires archiving of the trial master file for at least 25 years after the end of the trial, subject to any longer Union requirements. This period does not determine how long every specimen is to be retained. Article 28(2) CTR allows a separate request for consent to use data outside the protocol exclusively for scientific purposes.<a href="#_ftn28" id="_ftnref28">[28]</a></p>



<p>Access to the source records raises a further issue. Article 26(4) of Poland’s Act on Patients’ Rights permits disclosure of medical records to a higher education institution or research institute for scientific purposes without revealing information that identifies an individual. A company working with the university needs its own basis for access. Consortium partners should identify actual controller and processor roles, access to the key and all recipients. Processing likely to entail a high risk requires a data protection impact assessment. A transfer to a recipient outside the European Economic Area engages Chapter V GDPR.<a href="#_ftn29" id="_ftnref29">[29]</a></p>



<p>The European Health Data Space (EHDS) will facilitate secondary access to data. Most of Chapter IV will apply from 26 March 2029. Article 51(1)(q) covers data from biobanks, while the date for genetic, epigenomic and genomic data under point (f) is 26 March 2031. An applicant will need to specify the categories and a permitted purpose. If a biobank holds both clinical observations and genomic data, the two elements of the requested dataset need to be classified separately. The Regulation provides for data permits, a secure processing environment, an opt-out and a ban on use for advertising. To obtain pseudonymised data, a researcher must justify why anonymous data would not achieve the purpose. In 2026, citing the EHDS provides no immediate entitlement to a Polish cohort or to physical specimens.<a href="#_ftn30" id="_ftnref30">[30]</a></p>



<h1 class="wp-block-heading">Who can use the results and seek patent protection</h1>



<p>A joint university and biotechnology project needs to establish what each party brings to the research and what rights it can acquire. A laboratory may control the assay technique, a university the model code and a hospital the records and specimens entrusted to it for a limited purpose. A licence to use an algorithm alone confers no right to train it on another cohort, disclose data to an investor or submit them to a medicines authority. Rights to material and data must therefore be distinguished from rights to software and inventions, and from publication arrangements. The allocation should also allow the records needed for regulatory verification to be retained and made available.</p>



<p>Articles 5 and 6 of the Polish Clinical Trials Act impose a particular constraint. Under the ordinary definition of a non-commercial trial, the sponsor and any co-sponsors must fall within the statutory categories, which include universities, healthcare entities and investigators. The right to dispose of the trial data must belong exclusively to those entities. Article 5(1), point 3, excludes from this route trials undertaken as part of the scientific advice specified there or as part of a paediatric investigation plan. Article 6(1) restricts use of data to secure a marketing authorisation, to vary an authorisation or for marketing. It makes an exception for variations concerning the safety of the medicinal product. Article 6(2) and (3) prohibit transactions by the sponsor that enable the uses set out there and make contrary transactions void by operation of law.<a href="#_ftn31" id="_ftnref31">[31]</a></p>



<p>Article 6(4) disapplies paragraphs 1–3 to trials within Article 5(2). Such trials must be funded entirely from public funds and meet at least the requirements relating to eligible sponsors and control of the data. The mere existence of a grant therefore calls for further enquiry into the full budget, partner contributions and contractual terms. Article 7 permits free or discounted supply of the study medicine and scientific, technical or financial support from a manufacturer or marketing authorisation holder, provided the support is disclosed in the application and the relevant authorisation is obtained. Disclosure of such support does not in itself give that partner a right to use the data for registration. This matters when an investor seeks an option over the results before enrolment begins.<a href="#_ftn32" id="_ftnref32">[32]</a></p>



<p>For patent purposes, the discovery of a biological relationship must be distinguished from its technical application. A reagent kit, a new method of analysing a sample ex vivo or a defined further medical use of a known substance may be assessed. Novelty, inventive step and industrial application matter. For diagnostic methods, much turns on how the claims are framed and whether they encompass the required steps of diagnosis practised on the body. Publication disclosing the invention before a filing can destroy novelty. Consortium terms should allow time for filing while preserving the obligation to make trial results public.<a href="#_ftn33" id="_ftnref33">[33]</a></p>



<h1 class="wp-block-heading">Communicating a result within the limits of the evidence</h1>



<p>A participant report, a scientific abstract and a retail advertisement may cite the same number, but they invite different inferences. A paper can report the change on a named clock, a confidence interval, the comparator and disagreement between prespecified measures. On a package or sales page, “reverses ageing by three years” will ordinarily be understood as a statement about what the product does to a person. A footnote linking to a paper still leaves the advertiser to establish that the evidence supports the claim and that the advertising rules for the product permit it.</p>



<p>Polish Pharmaceutical Law prohibits, among other things, advertising medicines that are not authorised in Poland, using claims inconsistent with the summary of product characteristics in medicines advertising, and advertising prescription-only medicines to the public. For a test, Article 7 IVDR prohibits attributing functions or performance the device does not possess or suggesting uses outside the purpose covered by the conformity assessment. The Polish Medical Devices Act also requires public advertising to be understandable to a layperson and restricts public advertising of devices intended for professional users. It therefore matters whether a test campaign is directed to laboratories, prescribers or consumers.<a href="#_ftn34" id="_ftnref34">[34]</a></p>



<p>Food supplements are subject to EU health claims rules. A reference to general, non-specific health benefits must be accompanied by an authorised specific claim under Article 10(3) of Regulation (EC) No 1924/2006. As a rule, food information must not attribute disease prevention, treatment or cure to food. Claims about reducing a disease risk factor have a separate authorisation route under Article 14. For a new ingredient, the questions include whether it falls within a category in Article 3(2)(a) of Regulation (EU) 2015/2283, whether it was consumed to a significant degree in the EU before 15 May 1997, whether novel food authorisation is required and which conditions apply to its listing. Polish notification to the Chief Sanitary Inspector on first marketing a supplement is a market monitoring measure. It does not establish an anti-ageing effect.<a href="#_ftn35" id="_ftnref35">[35]</a></p>



<p>The practical value of an ageing biomarker is that a clinical question can be investigated before a distant patient outcome can be observed. In the opening example, a first paper can accurately report the difference in the clock score against the control. Using that score to decide who receives treatment requires validation of the test and the applicable regulatory route. Claiming that the therapy prolongs independent living requires evidence connecting its effect on the score with precisely that benefit. These decisions can be made in sequence. A programme designed on that basis preserves the value of the first observation while making clear what must be demonstrated next.</p>



<p data-pm-slice="1 1 []">KG LEGAL KIELTYKA GLADKOWSKI advises international life sciences and biotechnology businesses on Polish and EU regulatory requirements, research arrangements and commercialisation. Recognised by The Legal 500 for Healthcare and Life Sciences and Data Privacy and Data Protection, our firm combines sector knowledge with experience in corporate matters, negotiations and disputes.</p>
<p>We support projects where clinical research, technology and sensitive data intersect, helping clients address the legal requirements relevant to their research activities and business operations in Poland.</p>
<p>Contact: <a href="mailto:office@kg-legal.pl">office@kg-legal.pl</a></p>
<p>#LifeSciencesLaw #Biotechnology #ClinicalResearch #DataProtection #PolishLaw #KGLegal</p>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<p><a href="#_ftnref1" id="_ftn1">[1]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://www.ema.europa.eu/en/documents/scientific-guideline/ich-e-9-statistical-principles-clinical-trials-step-5_en.pdf">ICH E9</a>, Statistical Principles for Clinical Trials, section 2.2.6 (surrogate variables), EMA.</p>



<p><a href="#_ftnref2" id="_ftn2">[2]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://doi.org/10.1016/j.ebiom.2019.08.069">L.J. Hickson et al.</a>, Senolytics decrease senescent cells in humans, EBioMedicine 47 (2019), pp. 446–456.</p>



<p><a href="#_ftnref3" id="_ftn3">[3]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://eic.ec.europa.eu/system/files/2025-11/EIC-Work-Programme-2026.pdf">European Innovation Council</a>, EIC Work Programme 2026, section II.2.2, Biotechnology for Healthy Ageing, objective 2.</p>



<p><a href="#_ftnref4" id="_ftn4">[4]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://www.ema.europa.eu/en/documents/scientific-guideline/ich-e-9-statistical-principles-clinical-trials-step-5_en.pdf">ICH E9</a>, Statistical Principles for Clinical Trials, section 2.2.6 (surrogate variables), EMA.</p>



<p><a href="#_ftnref5" id="_ftn5">[5]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://doi.org/10.1038/s43587-022-00357-y">R. Waziry et al.</a>, Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial, Nature Aging 3 (2023), pp. 248–257.&nbsp; <a href="https://doi.org/10.1038/s41591-026-04562-9">R. Sehgal et al.</a>, Responsiveness of epigenetic aging biomarkers to longevity interventions in humans, Nature Medicine 32 (2026), pp. 3477–3490.</p>



<p><a href="#_ftnref6" id="_ftn6">[6]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://www.ema.europa.eu/en/documents/scientific-guideline/ich-e9-r1-addendum-estimands-and-sensitivity-analysis-clinical-trials-guideline-statistical-principles-clinical-trials-step-5_en.pdf">ICH E9(R1)</a>, Addendum on Estimands and Sensitivity Analysis in Clinical Trials, sections A.3–A.5, EMA.</p>



<p><a href="#_ftnref7" id="_ftn7">[7]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://www.ema.europa.eu/system/files/documents/regulatory-procedural-guideline/qualification-novel-methodologies-qonm-medicinal-product-development_en.pdf">EMA, Qualification of</a> novel methodologies for medicinal product development, procedural guidance for applicants, rev. 6 (2026).&nbsp; <a href="https://www.ema.europa.eu/en/human-regulatory-overview/research-development/scientific-advice-protocol-assistance">EMA, Scientific advice</a> and protocol assistance.</p>



<p><a href="#_ftnref8" id="_ftn8">[8]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02001L0083-20250101">Directive 2001/83/EC on</a> the Community code relating to medicinal products for human use, Articles 1(2) and 2(2).&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/612/ogl">Polish Pharmaceutical Law Act of 6 September 2001</a>, Article 2 point 32 and Article 3a, consolidated text, Journal of Laws 2026, item 612, as amended.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:62007CJ0140">CJEU</a>, Hecht-Pharma, C-140/07, judgment of 15 January 2009, EU:C:2009:5, paras 29, 32 and 39–42.</p>



<p><a href="#_ftnref9" id="_ftn9">[9]</a>&nbsp;&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02001L0083-20250101">Directive 2001/83/EC</a>, Articles 6, 8 and 26.&nbsp; <a href="https://www.ema.europa.eu/en/human-regulatory-overview/post-authorisation/variations-including-extensions-marketing-authorisations/extensions-marketing-authorisations-questions-answers">EMA, Extensions of</a> marketing authorisations: questions and answers, question 1.</p>



<p><a href="#_ftnref10" id="_ftn10">[10]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">Regulation (EU) No</a> 536/2014 on clinical trials on medicinal products for human use (CTR), Articles 1 and 2(2), points 1–4.</p>



<p><a href="#_ftnref11" id="_ftn11">[11]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">CTR, Article 2(2),</a> point 3.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32010L0063">Directive 2010/63/EU on</a> the protection of animals used for scientific purposes, Articles 4, 36 and 38.&nbsp; <a href="https://eli.gov.pl/eli/DU/2015/266/ogl">Polish Act of</a> 15 January 2015 on the Protection of Animals Used for Scientific or Educational Purposes, Articles 3a and 42.</p>



<p><a href="#_ftnref12" id="_ftn12">[12]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">CTR, Articles 4–8</a> and Annex I.&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/2/ogl/pol">Polish Act of</a> 9 March 2023 on Clinical Trials of Medicinal Products for Human Use, Articles 9 and 21–25, consolidated text, Journal of Laws 2026, item 2.</p>



<p><a href="#_ftnref13" id="_ftn13">[13]</a>&nbsp;&nbsp;&nbsp; <a href="https://www.ema.europa.eu/en/documents/scientific-guideline/reflection-paper-physical-frailty-instruments-baseline-characterisation-older-populations-clinical-trials-first-version_en.pdf">EMA/CHMP</a>, Reflection paper on physical frailty: instruments for baseline characterisation of older populations in clinical trials, EMA/CHMP/778709/2015, 9 January 2018, sections 5–6.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">CTR, Articles 28–31</a> and Annex I.</p>



<p><a href="#_ftnref14" id="_ftn14">[14]</a>&nbsp;&nbsp;&nbsp; <a href="https://pubmed.ncbi.nlm.nih.gov/40147475/">M.D. Witham et al.</a>, Metformin and physical performance in older people with probable sarcopenia and physical prefrailty or frailty in England (MET-PREVENT), The Lancet Healthy Longevity 6 (2025), 100695.</p>



<p><a href="#_ftnref15" id="_ftn15">[15]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">CTR, Articles 31,</a> 37, 41–43 and Annex V.&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/2/ogl/pol">Polish Clinical Trials Act</a>, Articles 39–44.</p>



<p><a href="#_ftnref16" id="_ftn16">[16]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0746">Regulation (EU) 2017/746</a> on in vitro diagnostic medical devices (IVDR), Articles 1(3)(a) and 2, points 2 and 12.&nbsp; <a href="https://health.ec.europa.eu/system/files/2022-05/mdcg_2022-10_en.pdf">MDCG 2022-10</a>, Q&amp;A on the interface between the CTR and the IVDR, May 2022, questions 5–7 and figure 1.</p>



<p><a href="#_ftnref17" id="_ftn17">[17]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0746">IVDR</a>, Article 2 point 7 and Article 48, Annex VIII, rule 3(f), and Annex IX, section 5.2.&nbsp; <a href="https://www.ema.europa.eu/en/human-regulatory-overview/medical-devices/companion-diagnostics-vitro-diagnostics">EMA, Companion diagnostics</a> (in vitro diagnostics).</p>



<p><a href="#_ftnref18" id="_ftn18">[18]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0746">IVDR</a>, Articles 56–58 and Annex XIII. See also MDCG 2022-10, question 7.</p>



<p><a href="#_ftnref19" id="_ftn19">[19]</a>&nbsp;&nbsp;&nbsp; <a href="https://www.gov.pl/web/urpl/badania-dzialania-wyrobow-medycznych-do-diagnostyki-in-vitro">URPL</a>, Performance studies of in vitro diagnostic medical devices (in Polish). See also the Polish Act of 7 April 2022 on Medical Devices, consolidated text, Journal of Laws 2024, item 1620, as amended.&nbsp; <a href="https://www.gov.pl/web/urpl/pytaniaiodpowiedzi">URPL</a>, Questions and answers, questions 4–5 on combined studies (in Polish).&nbsp; <a href="https://health.ec.europa.eu/document/download/183eb4ee-d093-4a6a-bf18-fa04a744448e_en?filename=md_combine_comb-proj1_sponsor-guide-phase2_en.pdf">European Commission</a>, COMBINE Project 1, phase two sponsor guide, sections 1–2 and Annex I.</p>



<p><a href="#_ftnref20" id="_ftn20">[20]</a>&nbsp;&nbsp;&nbsp; <a href="https://health.ec.europa.eu/document/download/b45335c5-1679-4c71-a91c-fc7a4d37f12b_en?filename=md_mdcg_2019_11_guidance_qualification_classification_software_en.pdf">MDCG 2019-11 rev. 1</a>, Guidance on Qualification and Classification of Software, June 2025.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0745">Regulation (EU) 2017/745 (MDR)</a>, Article 2 point 1 and Annex VIII, rule 11. See also IVDR, Article 2 point 2.</p>



<p><a href="#_ftnref21" id="_ftn21">[21]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689">Regulation (EU) 2024/1689 (AI Act)</a>, Articles 2(6) and (8), 3(1) and 6(1), and Annex I.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32026R1744">Regulation (EU) 2026/1744 (Digital Omnibus on AI)</a>, amendment to Article 113 of the AI Act, recital 38.</p>



<p><a href="#_ftnref22" id="_ftn22">[22]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2025/1295/ogl">Polish Act of</a> 15 September 2022 on Laboratory Medicine, consolidated text, Journal of Laws 2025, item 1295, as amended.&nbsp; <a href="https://eli.gov.pl/eli/DU/2025/961/ogl/pol">Polish Minister of</a> Health Regulation of 30 June 2025 on laboratory quality standards, Journal of Laws 2025, item 961, Annex 3.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0746">IVDR</a>, Article 4, in particular paragraphs 2–3.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32017R0746">IVDR</a>, Article 5(5), in particular points (a)–(c).</p>



<p><a href="#_ftnref23" id="_ftn23">[23]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/37/ogl/pol">Polish Act of</a> 5 December 1996 on the Professions of Physician and Dentist, Articles 21, 23, 23a, 23c, 25, 29 and 29a, consolidated text, Journal of Laws 2026, item 37, as amended.</p>



<p><a href="#_ftnref24" id="_ftn24">[24]</a>&nbsp;&nbsp;&nbsp; <a href="https://abm.gov.pl/download/1/9242/OdpowiedzinapytaniazczatuwebinariumTransmedI.pdf">Polish Medical Research Agency</a>, TRANSMED I webinar answers (ABM/2024/8), p. 1, archived and coded specimens (in Polish).&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/37/ogl/pol">Polish Act on</a> the Professions of Physician and Dentist, Article 24(2), points 5 and 7, and Articles 25 and 28.</p>



<p><a href="#_ftnref25" id="_ftn25">[25]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/37/ogl/pol">Polish Act on</a> the Professions of Physician and Dentist, Article 24(2), points 5 and 7, and Articles 25 and 28.</p>



<p><a href="#_ftnref26" id="_ftn26">[26]</a>&nbsp;&nbsp;&nbsp; <a href="https://www.nik.gov.pl/komunikaty/badania-genetyczne-moga-zmienic-zycie-calej-rodziny-panstwo-od-lat-nie-ustalilo-kompleksowych-zasad-ochrony-materialu-genetycznego">Polish Supreme Audit Office (NIK)</a>, findings on genetic testing and safeguards for genetic material, 22 October 2026 (in Polish).</p>



<p><a href="#_ftnref27" id="_ftn27">[27]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32016R0679">Regulation (EU) 2016/679 (GDPR)</a>, Article 4, points 5, 13 and 15, Articles 5, 6, 9 and 89, and recital 26.</p>



<p><a href="#_ftnref28" id="_ftn28">[28]</a>&nbsp;&nbsp;&nbsp; <a href="https://www.edpb.europa.eu/documents/legislative-opinion/opinion-32019-concerning-the-questions-and-answers-on-the-interplay_en">EDPB</a>, Opinion 3/2019 on the interplay between the CTR and the GDPR, 23 January 2019, sections 2–3.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32014R0536">CTR, Articles 28(2)–(3)</a> and 58.</p>



<p><a href="#_ftnref29" id="_ftn29">[29]</a>&nbsp;&nbsp;&nbsp; <a href="https://api.sejm.gov.pl/eli/acts/DU/2024/581/text.html">Polish Act of</a> 6 November 2008 on Patients’ Rights and the Commissioner for Patients’ Rights, Article 26(4), consolidated text, Journal of Laws 2024, item 581, as amended.&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32016R0679">GDPR</a>, Articles 26, 28, 32 and 35, and Chapter V.</p>



<p><a href="#_ftnref30" id="_ftn30">[30]</a>&nbsp;&nbsp;&nbsp; <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32025R0327">Regulation (EU) 2025/327</a> on the European Health Data Space, Articles 51, 53–54, 66, 68, 71, 73 and 105.</p>



<p><a href="#_ftnref31" id="_ftn31">[31]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/2/ogl/pol">Polish Clinical Trials Act</a>, Articles 5(1) and 6(1)–(3).</p>



<p><a href="#_ftnref32" id="_ftn32">[32]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/2/ogl/pol">Polish Clinical Trials Act</a>, Articles 5(2), 6(4) and 7.</p>



<p><a href="#_ftnref33" id="_ftn33">[33]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/api/acts/DU/2023/1170/text.html">Polish Industrial Property Act of 30 June 2000</a>, Articles 24–29, consolidated text, Journal of Laws 2023, item 1170, as amended.&nbsp; <a href="https://www.epo.org/en/legal/guidelines-epc/2026/g_ii_4_2_1_3.html">European Patent Office</a>, Guidelines for Examination 2026, G-II, 4.2.1.3, referring to G 1/04 and Article 53(c) of the European Patent Convention.&nbsp; <a href="https://www.epo.org/en/legal/guidelines-epc/2026/g_vi_6_1.html">European Patent Office</a>, Guidelines for Examination 2026, G-VI, 6.1, Article 54(4)–(5) EPC.</p>



<p><a href="#_ftnref34" id="_ftn34">[34]</a>&nbsp;&nbsp;&nbsp; <a href="https://eli.gov.pl/eli/DU/2026/612/ogl">Polish Pharmaceutical Law</a>, Articles 52–57, in particular Articles 56 and 57(1).&nbsp; <a href="https://eli.gov.pl/eli/DU/2024/1620/ogl">Polish Medical Devices Act</a>, Articles 54–55, read with Article 7 IVDR.</p>



<p><a id="_ftn35" href="#_ftnref35">[35]</a>    <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32006R1924">Regulation (EC) No</a> 1924/2006 on nutrition and health claims made on foods, Articles 2, 10 and 14.  <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32011R1169">Regulation (EU) No</a> 1169/2011 on food information to consumers, Article 7(3)–(4).  <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32015R2283">Regulation (EU) 2015/2283 on novel foods</a>, Articles 3, 6 and 10.  <a href="https://eli.gov.pl/eli/DU/2026/1399/ogl">Polish Food and</a> Nutrition Safety Act of 25 August 2006, Article 29, consolidated text, Journal of Laws 2026, item 1399.</p>
<p> </p>



<div class="wp-block-buttons is-layout-flex wp-block-buttons-is-layout-flex"></div>






<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ageing-biomarkers-as-evidence-of-geroprotective-efficacy/">Ageing biomarkers as evidence of geroprotective efficacy</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ageing-biomarkers-as-evidence-of-geroprotective-efficacy/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Poland updates lawful interception interface rules</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/poland-updates-lawful-interception-interface-rules/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/poland-updates-lawful-interception-interface-rules/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Tue, 10 Nov 2026 15:50:36 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[5G Networks]]></category>
		<category><![CDATA[Agency Agreements]]></category>
		<category><![CDATA[ASN.1]]></category>
		<category><![CDATA[Compliance Costs]]></category>
		<category><![CDATA[Data Access]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[Electronic Communications Law]]></category>
		<category><![CDATA[Electronic Signatures]]></category>
		<category><![CDATA[ETSI Standards]]></category>
		<category><![CDATA[Fixed Network Operators]]></category>
		<category><![CDATA[HI A-B]]></category>
		<category><![CDATA[Interface Migration]]></category>
		<category><![CDATA[Internet Service Providers]]></category>
		<category><![CDATA[Law Enforcement Access]]></category>
		<category><![CDATA[Lawful Interception]]></category>
		<category><![CDATA[Lawful Interception Interfaces]]></category>
		<category><![CDATA[LI HI]]></category>
		<category><![CDATA[Mobile Operators]]></category>
		<category><![CDATA[Poland]]></category>
		<category><![CDATA[Polish law]]></category>
		<category><![CDATA[Polish Telecommunications Law]]></category>
		<category><![CDATA[Public Key Infrastructure]]></category>
		<category><![CDATA[Real-Time Data Delivery]]></category>
		<category><![CDATA[Regulatory Compliance]]></category>
		<category><![CDATA[Retained Data]]></category>
		<category><![CDATA[Security Documentation]]></category>
		<category><![CDATA[TCP]]></category>
		<category><![CDATA[Telecom Due Diligence]]></category>
		<category><![CDATA[Telecom Regulation]]></category>
		<category><![CDATA[Telecommunications]]></category>
		<category><![CDATA[Telecommunications M&A]]></category>
		<category><![CDATA[Telecommunications Security]]></category>
		<category><![CDATA[Transitional Provisions]]></category>
		<category><![CDATA[Vendor Contracts]]></category>
		<category><![CDATA[VoIP Providers]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8983</guid>

					<description><![CDATA[<p>Publication date: November 10, 2026 10 November 2026 From 11 November 2026, telecommunications undertakings in Poland must meet new technical and operational requirements for the interfaces that give law enforcement and security agencies access to communications and related data. Interfaces already running under existing agreements may stay on the previous standard, but for no longer [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/poland-updates-lawful-interception-interface-rules/">Poland updates lawful interception interface rules</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color"><strong>Publication date: November 10, 2026</strong></mark></p>



<p>10 November 2026</p>



<p><strong>From 11 November 2026, telecommunications undertakings in Poland must meet new technical and operational requirements for the interfaces that give law enforcement and security agencies access to communications and related data. Interfaces already running under existing agreements may stay on the previous standard, but for no longer than 36 months. The new rules add 5G standards, real-time delivery and a dedicated interface for retained data.</strong></p>



<span id="more-8983"></span>



<h2 class="wp-block-heading">Who is affected</h2>



<p>The rules concern telecommunications undertakings that perform their statutory duties through an interface agreed with the authorised agencies. This covers mobile and fixed operators, internet access providers and VoIP providers.</p>



<p>The duty itself comes from the Electronic Communications Law of 12 July 2024. Article 43 requires an undertaking to provide, at its own cost, the technical and organisational conditions for access to and recording of communications and related data. Article 45 adds a duty to prepare the conditions for making processed data available.</p>



<p>An interface is one permitted way of meeting those duties, not the only one. Where it is used, its operation is governed by an agreement between the undertaking and the agencies. That agreement must at least allocate interface costs and set a procedure for failures. Micro and small undertakings perform the duty in proportion to their scale.</p>



<p>The regulation lists nine authorised agencies: the Internal Security Agency (ABW), the Police, the Military Counterintelligence Service (SKW), the Military Gendarmerie, the Border Guard, the National Revenue Administration (KAS), the Central Anticorruption Bureau (CBA), the State Protection Service (SOP) and the Internal Supervision Bureau (BNW).</p>



<h2 class="wp-block-heading">Two interfaces, two purposes</h2>



<p>The Council of Ministers regulation of 29 October 2026 sets requirements for two IP-based interfaces between the operator&#8217;s system and the agencies&#8217; monitoring systems.</p>



<figure class="wp-block-table"><table class="has-fixed-layout"><thead><tr><td><strong>Interface</strong></td><td><strong>Purpose</strong></td><td><strong>Components</strong></td></tr></thead><tbody><tr><td>LI HI (lawful interception)</td><td>Access to communications and related data under Article 43</td><td>HI1: orders and status messages. HI2: intercept related information and SMS content. HI3: content of communications</td></tr><tr><td>HI A-B (retained data)</td><td>Delivery of data under Article 45</td><td>HI A: requests and their handling. HI B: results of requests</td></tr></tbody></table></figure>



<p>Annex 1 contains the new requirements. Annex 2 preserves the previous standard for interfaces covered by the transition period.</p>



<h2 class="wp-block-heading">What changes for lawful interception</h2>



<p>The new standard moves from file transfer to streamed, real-time delivery and brings 5G into scope.</p>



<figure class="wp-block-table"><table class="has-fixed-layout"><thead><tr><td><strong>Area</strong></td><td><strong>Previous standard (Annex 2)</strong></td><td><strong>New standard (Annex 1)</strong></td></tr></thead><tbody><tr><td>Intercept related information</td><td>Files sent over FTP</td><td>TCP streams under ETSI TS 102 232-1</td></tr><tr><td>Content of communications in mobile networks</td><td>Offline files over FTP, or online mode over SIP</td><td>Real-time delivery over TCP, no separate online mode</td></tr><tr><td>5G</td><td>No 5G standard listed</td><td>ETSI TS 133 128 added</td></tr><tr><td>Buffering</td><td>No fixed period</td><td>72 hours</td></tr><tr><td>Versions of standards</td><td>Fixed versions</td><td>Newer versions allowed, and other versions if they perform the duties to no lesser degree</td></tr><tr><td>Codecs</td><td>Supplied to agencies free of charge</td><td>Agency may request them, no reference to cost</td></tr><tr><td>Physical layer</td><td>Ethernet</td><td>Ethernet, or fibre if both sides agree</td></tr></tbody></table></figure>



<p>Several core rules carry over unchanged:</p>



<ul class="wp-block-list">
<li>Intercept related information must reach the agency no later than 10 minutes after the communication ends.</li>



<li>Orders are executed automatically, without the operator&#8217;s staff.</li>



<li>A semi-automatic mode needs the agency&#8217;s consent and gives authorised staff up to 24 hours to prepare the network.</li>



<li>Activation, modification and deactivation orders carry an electronic signature.</li>



<li>Data is deleted from the operator&#8217;s resources once delivered.</li>
</ul>



<h2 class="wp-block-heading">A 36-month transition, with limits</h2>



<p>Existing lawful interception interfaces may keep to the previous standard until November 2029 at the latest. The relief has three conditions:</p>



<ol class="wp-block-list">
<li>It covers LI HI interfaces only.</li>



<li>The interface must operate under an agreement concluded before 11 November 2026.</li>



<li>It applies only to the telecommunications services covered by that agreement.</li>
</ol>



<p>A new agreement, or a new service added outside an existing one, falls under the new standard from the start. Operators planning 5G standalone launches or new VoIP products should check which side of the line each service sits on.</p>



<h2 class="wp-block-heading">Retained data: no transition period</h2>



<p>The transitional rule does not mention the HI A-B interface, so the new requirements apply to it from day one. The interface must follow ETSI TS 102 657, use TCP with ASN.1/BER encoding and require an electronic signature. It must be available at a single location for all agencies, with a separate Ethernet port for each.</p>



<p>Some optional features of the ETSI standard are switched off: request priorities, multi-part delivery and the mode in which the agency initiates delivery.</p>



<p>In practice, operators should confirm with the agencies whether and when they expect this interface to be used. During public consultation, an industry chamber stated that neither side had deployed HI A-B under the earlier rules.</p>



<h2 class="wp-block-heading">Privacy points worth noting</h2>



<p>The regulation is technical, but several provisions limit what the operator keeps and discloses:</p>



<ul class="wp-block-list">
<li><strong>Deletion after delivery. </strong>The operator removes intercepted communications and related data once they are successfully handed over.</li>



<li><strong>Limited buffer. </strong>Undelivered data is held for 72 hours.</li>



<li><strong>Excluded data fields. </strong>The retained data interface does not carry date of birth, gender, identification number, authentication data, profession or telephony billing details.</li>



<li><strong>No identifying data in status replies. </strong>A reply to a status query contains only the interception identifier.</li>



<li><strong>Signed orders. </strong>Interception orders must carry an electronic signature based on certificates issued within a public key infrastructure.</li>
</ul>



<p>These points belong in the operator&#8217;s records of processing and security documentation.</p>



<h2 class="wp-block-heading">What operators should do now</h2>



<ol class="wp-block-list">
<li><strong>Map existing agreements. </strong>List every agreement with an authorised agency, its date and the services it covers. This decides which interfaces qualify for the transition.</li>



<li><strong>Plan the migration. </strong>Set a schedule to move legacy interfaces to the new standard before November 2029, with time for testing with each agency.</li>



<li><strong>Review vendor contracts. </strong>Check whether suppliers of interception and mediation systems are bound to deliver compliance with the new ETSI versions, and who pays.</li>



<li><strong>Revisit cost and failure clauses. </strong>Agreements with agencies must allocate interface costs and set a failure procedure. The 72-hour buffer and the codec rule are natural points to renegotiate.</li>



<li><strong>Clarify the retained data interface. </strong>Ask the agencies about their plans for HI A-B, since no transition period applies.</li>



<li><strong>Include the topic in transactions. </strong>Buyers of Polish telecommunications businesses should treat interface compliance and migration cost as a due diligence item.</li>
</ol>



<p>KG Legal KIELTYKA GLADKOWSKI advises telecommunications and technology companies on Polish and EU regulatory requirements. Contact: office@kg-legal.pl.</p>
<p> </p>


<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/poland-updates-lawful-interception-interface-rules/">Poland updates lawful interception interface rules</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/poland-updates-lawful-interception-interface-rules/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Paying Interest on Bad Code: Why Technical Debt Has Become a Legal Problemh</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/technical-debt-and-legal-risk-gdpr-and-nis2-explained/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/technical-debt-and-legal-risk-gdpr-and-nis2-explained/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Wed, 04 Nov 2026 11:27:57 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[AI Accountability]]></category>
		<category><![CDATA[Architecture Decision Records]]></category>
		<category><![CDATA[ASVS]]></category>
		<category><![CDATA[Automated Testing]]></category>
		<category><![CDATA[CI/CD]]></category>
		<category><![CDATA[cross border cases]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[Doing business in Poland]]></category>
		<category><![CDATA[Dynamic Application Security Testing]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[GDPR Article 32]]></category>
		<category><![CDATA[ISO/IEC 27001]]></category>
		<category><![CDATA[IT Governance]]></category>
		<category><![CDATA[IT Law]]></category>
		<category><![CDATA[KG Legal]]></category>
		<category><![CDATA[kglegal]]></category>
		<category><![CDATA[Legacy Code]]></category>
		<category><![CDATA[Legacy Systems]]></category>
		<category><![CDATA[Legal Risk]]></category>
		<category><![CDATA[Management Liability]]></category>
		<category><![CDATA[NIS2]]></category>
		<category><![CDATA[OWASP]]></category>
		<category><![CDATA[Patch Management]]></category>
		<category><![CDATA[Poland]]></category>
		<category><![CDATA[Polish KSC Act]]></category>
		<category><![CDATA[Polish law]]></category>
		<category><![CDATA[Refactoring]]></category>
		<category><![CDATA[Regulatory Compliance]]></category>
		<category><![CDATA[Secure Software Development]]></category>
		<category><![CDATA[Service Level Agreements]]></category>
		<category><![CDATA[Software Composition Analysis]]></category>
		<category><![CDATA[Software Documentation]]></category>
		<category><![CDATA[Software Maintenance]]></category>
		<category><![CDATA[Software Quality]]></category>
		<category><![CDATA[Software Supply Chain Security]]></category>
		<category><![CDATA[Static Application Security Testing]]></category>
		<category><![CDATA[Technical Debt]]></category>
		<category><![CDATA[Test-Driven Development]]></category>
		<category><![CDATA[Third-Party Dependencies]]></category>
		<category><![CDATA[Vulnerability Management]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8978</guid>

					<description><![CDATA[<p>Publication date: November 04, 2026 Technical debt, GDPR and NIS2 – how yesterday’s shortcuts turn into tomorrow’s liability Why technical debt arises Technical debt is a phenomenon that sooner or later affects almost every IT project. It consists in choosing a solution that is quicker to deliver but inferior in technical quality. A team usually [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/technical-debt-and-legal-risk-gdpr-and-nis2-explained/">Paying Interest on Bad Code: Why Technical Debt Has Become a Legal Problemh</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: November 04, 2026</mark></strong></p>



<p><em>Technical debt, GDPR and NIS2 – how yesterday’s shortcuts turn into tomorrow’s liability</em></p>



<h2 class="wp-block-heading">Why technical debt arises</h2>



<p>Technical debt is a phenomenon that sooner or later affects almost every IT project. It consists in choosing a solution that is quicker to deliver but inferior in technical quality. A team usually “takes out” this loan under pressure from the business side, for which the key metric is <strong>time-to-market</strong> – the sooner the product reaches the market, the better.</p>



<p>In the short term this brings a quick win, but in the future the company has to start paying the “interest”. It shows up as ever higher system maintenance costs, a slowdown in the pace of work, and the mounting frustration of developers who, instead of developing the product, must keep fixing the same defects over and over. Although the term “technical debt” appears nowhere in legislation, concepts such as <strong>due diligence</strong> and the <strong>adequacy of technical measures</strong> in practice force companies to keep that debt from spiralling out of control.</p>



<span id="more-8978"></span>



<h2 class="wp-block-heading">The forms that technological neglect takes</h2>



<p>Technical debt is not homogeneous; it accumulates in many places. One of the biggest problems is <strong>legacy code</strong> – code built on technologies that are no longer maintained. It often degenerates into so-called <strong>spaghetti code</strong>: a structure so tangled that a change in one place causes an unexpected failure in an entirely different part of the system.</p>



<p>Another barrier is the absence of automated tests, which forces the team into manual testing – where mistakes are easiest to make. Closely related is <strong>code duplication</strong>, that is, solving the same problem repeatedly in different parts of the application. Nor should the lack of documentation be forgotten, which gives rise to so-called <strong>tribal knowledge</strong>. When key employees leave, the organisation is left with a system that no one can operate or develop efficiently.</p>



<h2 class="wp-block-heading">The consequences for business and cybersecurity</h2>



<p>Technical debt is not merely an IT department problem – it strikes directly at the stability of the company. Outdated architecture leads to outages and performance degradation, which can result in breaches of <strong>SLA</strong> commitments and financial penalties. The greatest nightmare, however, are <strong>regressions</strong>: situations in which new features break functionality that previously worked correctly.</p>



<p>The most serious risks, though, are those to security. Failure to update components (<strong>third-party dependencies</strong>) exposes the company to <strong>zero-day</strong> attacks and to vulnerabilities catalogued in public <strong>CVE</strong> databases. Hackers routinely scan the internet for systems running old libraries for which security patches are no longer issued. Ignoring this problem is the simplest way to become an easy target.</p>



<h2 class="wp-block-heading">Technical debt under the GDPR and the “state of the art” principle</h2>



<p>The law today requires companies to take a proactive approach to risk. Under the <strong>GDPR</strong> (Article 32), controllers must implement measures that take account of the current <strong>state of the art</strong>. Using outdated encryption methods, or libraries that no longer receive updates, may amount to a breach of these provisions.</p>



<p>Technical debt also strikes at the principle of <strong>accountability</strong>. If, because of the poor state of its code, a company is unable to give effect efficiently to, say, the “right to be forgotten”, the supervisory authority may treat this as gross negligence. In the event of a data breach, the existence of substantial technical debt will be an aggravating circumstance – and may drastically increase the amount of the fine.</p>



<h2 class="wp-block-heading">Management board liability: NIS2 and the Polish KSC Act</h2>



<p>New regulations – the <strong>NIS2 Directive</strong> and the amended Polish <strong>Act on the National Cybersecurity System (KSC)</strong> implementing it, in force since 3 April 2026 – place responsibility for these matters directly on senior management. NIS2 puts strong emphasis on supply chain security and vulnerability management. If a system is so fragile that installing a patch risks bringing it down, the management board is caught in a trap: the law requires vulnerabilities to be patched, while the technical condition of the system makes that impossible.</p>



<p>Such omissions carry not only financial penalties but even a ban on holding managerial positions. Standards such as <strong>ISO/IEC 27001</strong>, <strong>ASVS</strong> and the <strong>OWASP</strong> guidelines are becoming the points of reference – ignoring them is treated as knowingly accepting an excessive level of risk.</p>



<h1 class="wp-block-heading">Strategies and prevention</h1>



<p>Managing technical debt does not mean halting all work; it means introducing healthy rules. Key among them is the <strong>boy scout rule</strong> – leave the code in a better state than you found it. It is also worth applying <strong>TDD (test-driven development)</strong>, that is, writing tests before the code, which builds a “safety net”.</p>



<p>To combat vulnerabilities, <strong>SCA</strong>-class tools that automatically monitor external packages are indispensable. <strong>CI/CD</strong> pipelines with built-in static (<strong>SAST</strong>) and dynamic (<strong>DAST</strong>) analysis should be the standard. To stop knowledge leaking out of the company, it is worth adopting a <strong>docs-as-code</strong> model (documentation kept together with the code) and maintaining architecture decision records (<strong>ADRs</strong>). In this way it is possible to build a system that is not only modern but, above all, secure and legally compliant.</p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/technical-debt-and-legal-risk-gdpr-and-nis2-explained/">Paying Interest on Bad Code: Why Technical Debt Has Become a Legal Problemh</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/technical-debt-and-legal-risk-gdpr-and-nis2-explained/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>Liability for Loss and Corruption of Data</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/liability-for-data-loss-and-corruption/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/liability-for-data-loss-and-corruption/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Mon, 05 Oct 2026 12:28:22 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[cloud services]]></category>
		<category><![CDATA[contractual liability]]></category>
		<category><![CDATA[data corruption]]></category>
		<category><![CDATA[data loss]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[data security]]></category>
		<category><![CDATA[disaster recovery]]></category>
		<category><![CDATA[DORA]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[IT contracts]]></category>
		<category><![CDATA[IT liability]]></category>
		<category><![CDATA[liability for data loss]]></category>
		<category><![CDATA[NIS2]]></category>
		<category><![CDATA[non-personal data]]></category>
		<category><![CDATA[personal data]]></category>
		<category><![CDATA[RPO]]></category>
		<category><![CDATA[RTO]]></category>
		<category><![CDATA[SLA]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8951</guid>

					<description><![CDATA[<p>Publication date: October 05, 2026 The loss or corruption of data in a business is rarely &#8220;just&#8221; a technical problem. In practice it is a serious legal and financial crisis: production downtime, loss of trust among business partners and, in extreme cases, administrative fines running into millions. The scope of liability, however, depends above all [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/liability-for-data-loss-and-corruption/">Liability for Loss and Corruption of Data</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: October 05, 2026</mark></strong></p>



<p><em>The loss or corruption of data in a business is rarely &#8220;just&#8221; a technical problem. In practice it is a serious legal and financial crisis: production downtime, loss of trust among business partners and, in extreme cases, administrative fines running into millions. The scope of liability, however, depends above all on what kind of data has been lost or corrupted. The law treats the loss of source code or technical documentation quite differently from a leak of employee records or a customer database. The first step in assessing liability for a breach of the integrity or availability of data is therefore its unambiguous legal classification.</em></p>



<span id="more-8951"></span>



<h1 class="wp-block-heading">Personal Data and Non-Personal Data</h1>



<p>It is worth starting with the category that, in recent years, has come under particular legal protection as a result of stringent EU rules: personal data, the protection of which directly concerns the privacy and rights of natural persons. Personal data protection is governed by Regulation (EU) 2016/679 of the European Parliament and of the Council on the protection of natural persons with regard to the processing of personal data and on the free movement of such data (GDPR). Under Article 4(1) GDPR, personal data means any information relating to an identified or identifiable natural person, i.e. a person who can be identified, directly or indirectly, in particular by reference to an identifier such as a name or a national identification number (in Poland, the PESEL number).</p>



<p>The rapid growth of the data-driven economy means that non-personal data is gaining ever greater economic and legal significance. Non-personal data is data that does not relate to an identified or identifiable natural person and therefore falls outside the scope of the GDPR. As the recitals of Regulation (EU) 2018/1807 of the European Parliament and of the Council on a framework for the free flow of non-personal data point out, information and communication technologies are no longer a separate sector but the foundation of all modern economic systems, with electronic data at their core. The statutory definition of non-personal data is contained in Article 3(1) of Regulation 2018/1807 and Article 2(4) of Regulation 2022/868 (the Data Governance Act): it is data other than personal data within the meaning of the GDPR. This category includes in particular:</p>



<ul class="wp-block-list">
<li>technical, operational, industrial and telemetry data,</li>



<li>financial and production data,</li>



<li>source code and design documentation,</li>



<li>artificial intelligence models together with their training data,</li>



<li>data on the operation of IT systems.</li>
</ul>



<p>Non-personal data includes both data that has never contained any information about natural persons (e.g. weather data or share prices) and data that was originally personal but has subsequently been effectively anonymised, i.e. to a degree that permanently prevents the identification of a natural person. It should be borne in mind that data which has merely been pseudonymised remains personal data (recital 26 GDPR). Non-personal data may be a key asset of a business, often worth more than its traditional tangible assets; in the case of artificial intelligence models, the quality, volume and diversity of the data directly determine the value and effectiveness of the model itself.</p>



<h2 class="wp-block-heading">A Different Centre of Gravity of Protection</h2>



<p>The fundamental distinction between personal and non-personal data is not merely a technical or definitional matter. It determines where the centre of gravity of legal protection lies:</p>



<ul class="wp-block-list">
<li>for personal data, the central protected value is the rights and freedoms of natural persons, including the right to privacy and the right to informational self-determination; the GDPR regime is therefore constructed primarily from the perspective of the data subject as the weaker party in need of protection;</li>



<li>for non-personal data, the centre of gravity shifts dramatically: what matters is the economic interest of the business, its operational continuity, organisational security and the economic value of the data as such.</li>
</ul>



<p>The loss or corruption of non-personal data does not infringe the rights of any natural person within the meaning of the GDPR, but it may lead to serious financial losses, loss of competitive advantage, disruption of production processes or the permanent destruction of a resource that cannot be recreated at all, or only at disproportionate cost. This distinction translates into different bases of liability, because non-personal data is not subject to the sanctions provided for in the GDPR.</p>



<h2 class="wp-block-heading">Mixed Data Sets</h2>



<p>A separate issue is that of mixed data sets, i.e. data sets composed of both personal and non-personal data (e.g. the ERP and CRM systems of online shops). Under Article 2(2) of Regulation 2018/1807, where personal and non-personal data are inextricably linked, the GDPR applies to the whole data set. In other cases the GDPR applies only to the personal data part of the set, while non-personal data is governed primarily by the rules of contractual liability arising from the agreement and by the provisions of civil law.</p>



<figure class="wp-block-table"><table class="has-fixed-layout"><thead><tr><td><strong>Criterion</strong></td><td><strong>Personal data</strong></td><td><strong>Non-personal data</strong></td></tr></thead><tbody><tr><td>Object of protection</td><td>Rights and freedoms of the natural person</td><td>Economic, organisational and technological interest</td></tr><tr><td>Main legal regime</td><td>GDPR</td><td>Civil law, contracts, sectoral regulation, cybersecurity law</td></tr><tr><td>Typical harm</td><td>Identity theft, distress, material or non-material damage</td><td>Downtime, loss of data value, wrong business decisions, contractual loss</td></tr><tr><td>Liable party</td><td>Controller, processor</td><td>IT provider, counterparty, management board, cloud service provider</td></tr><tr><td>Typical instruments</td><td>DPIA, breach notification, communication to the data subject</td><td>SLA, backup, RPO/RTO, disaster recovery, security audit, notification of the authority</td></tr><tr><td>Main practical problem</td><td>Exercise of the data subject&#8217;s rights</td><td>Recovery of the data and proof of economic loss</td></tr></tbody></table></figure>



<h1 class="wp-block-heading">Liability for Loss or Corruption of Personal Data</h1>



<p>To assess the consequences of the loss or corruption of personal data properly, three distinct concepts must be kept apart:</p>



<ul class="wp-block-list">
<li>a personal data breach (the security incident),</li>



<li>an infringement of the GDPR by the controller or processor,</li>



<li>damage suffered by the data subject.</li>
</ul>



<p>The occurrence of an incident does not automatically mean that the law has been infringed, and an infringement of the law does not in itself give rise to a right to compensation.</p>



<h2 class="wp-block-heading">Personal Data Breach</h2>



<p>Under Article 4(12) GDPR, a personal data breach is a breach of security leading to the accidental or unlawful destruction, loss, alteration, unauthorised disclosure of, or access to, personal data transmitted, stored or otherwise processed. Both the loss and the corruption of data therefore fall within the broad scope of this concept and may give rise to legal liability. A personal data breach may concern:</p>



<ul class="wp-block-list">
<li>confidentiality &#8211; unauthorised disclosure of, or access to, the data;</li>



<li>integrity &#8211; unauthorised alteration of the data, i.e. any change made by an unauthorised person or an incorrect change made by an authorised person;</li>



<li>availability &#8211; unauthorised loss of data (a situation in which the data cannot be used, temporarily or permanently, although it can be recovered or recreated) or unauthorised destruction of data (the data is lost irretrievably because the controller has no means of recreating it).</li>
</ul>



<p>A breach may occur by accident or as a result of deliberate and unlawful action. The most common causes include:</p>



<ul class="wp-block-list">
<li>human error, e.g. a mistake, a lost storage device or a failure to recognise that a breach has occurred;</li>



<li>inadequate safeguards or procedures;</li>



<li>cybercrime, e.g. phishing (impersonating a trusted person in order to extract sensitive data) or ransomware (an attack that blocks access to a system or renders data unreadable, combined with a demand for payment to restore the original state);</li>



<li>physical or environmental factors, such as natural disasters and failures of technical infrastructure.</li>
</ul>



<p>A personal data breach occurs regardless of whether any adverse consequences actually materialise. A failure to respond appropriately and promptly, however, may result in adverse consequences for the data subject, such as physical harm, material or non-material damage, or identity theft or fraud (recital 85 GDPR).</p>



<h2 class="wp-block-heading">Infringement of the GDPR</h2>



<p>Article 5(1)(f) GDPR sets out the principle of integrity and confidentiality: personal data must be processed in a manner that ensures appropriate security, including protection against unauthorised or unlawful processing and against accidental loss, destruction or damage, using appropriate technical or organisational measures. Article 5(2) GDPR introduces the principle of accountability, under which the controller must be able to demonstrate compliance with these principles.</p>



<p>The GDPR imposes on controllers a range of obligations relating to secure processing and to responding to breaches. These obligations are built on a risk-based approach: protective measures must be tailored to the risk that the processing poses to the rights and freedoms of natural persons. The requirements therefore differ from controller to controller depending on the specifics of the processing, and controllers must assess the associated risk themselves. The key obligations are:</p>



<ul class="wp-block-list">
<li>Article 24(1) GDPR &#8211; the obligation to implement appropriate technical and organisational measures and to assess risk;</li>



<li>Article 25(1) GDPR &#8211; the obligation to take data protection into account already at the design stage of the processing and throughout the processing itself (privacy by design);</li>



<li>Article 32(1) GDPR &#8211; the obligation to implement measures ensuring a level of security appropriate to the risk, e.g. authentication procedures and appropriate infrastructure and safeguards;</li>



<li>Articles 33 and 34 GDPR &#8211; the obligation to notify a breach to the supervisory authority (as a rule within 72 hours) and, where the breach is likely to result in a high risk to the rights and freedoms of natural persons, also to communicate it to the data subject;</li>



<li>Article 35 GDPR &#8211; the obligation to carry out a formal data protection impact assessment (DPIA) where a type of processing is likely to result in a high risk (e.g. large-scale processing of special categories of data or of data relating to criminal convictions); this obligation is further specified by the list of processing operations published by the President of the Polish Personal Data Protection Office (PUODO).</li>
</ul>



<p>Risk assessment consists in estimating the severity of the potential consequences of a breach and the likelihood of their occurrence, taking into account, among other things, the type of breach, the sensitivity of the data and the seriousness of the consequences for the data subjects.</p>



<p>An infringement of the GDPR is therefore conduct that fails to meet the above requirements. It is legal in nature, as opposed to the incident itself, which is a matter of fact. Failure to comply with the obligations listed in Articles 8, 11, 25 to 39, 42 and 43 GDPR is punishable under Article 83(4) GDPR by an administrative fine of up to EUR 10,000,000 or, in the case of an undertaking, up to 2% of its total worldwide annual turnover of the preceding financial year, whichever is higher.</p>



<p>The Regulation also grants the data subject a number of rights, the exercise of which may become difficult or impossible once data has been lost or corrupted:</p>



<ul class="wp-block-list">
<li>the right of access (Article 15 GDPR) &#8211; the right to obtain confirmation as to whether the controller is processing personal data and, if so, access to that data; after loss or corruption, the controller may be unable to give effect to this right, and if it fails to act on the request within one month (Article 12(3) and (4) GDPR), the data subject may lodge a complaint with the supervisory authority and seek a judicial remedy;</li>



<li>the right to rectification (Article 16 GDPR) &#8211; of particular importance where data has been corrupted; the difficulty may be that the controller does not know which data has been corrupted or to what state it should be restored, e.g. where the corruption has also affected the backup copies;</li>



<li>the right to erasure (Article 17 GDPR) &#8211; inter alia where the data is no longer necessary for the purposes for which it was collected, the data subject has withdrawn consent and there is no other legal basis, or the data has been processed unlawfully; where data has been lost (as opposed to destroyed) it still exists and can be restored, and a controller that cannot locate it will be unable to comply with an erasure request, which may in itself constitute an infringement of the GDPR;</li>



<li>the right to restriction of processing (Article 18(1)(a) GDPR) &#8211; where the data subject contests the accuracy of the data, which is precisely the case when data has been corrupted.</li>
</ul>



<p>Infringement of the provisions governing data subjects&#8217; rights (Articles 12 to 22 GDPR) and of the basic principles of processing (Articles 5 to 7 and 9 GDPR) is punishable under Article 83(5) GDPR by an administrative fine of up to EUR 20,000,000 or, in the case of an undertaking, up to 4% of its total worldwide annual turnover of the preceding financial year, whichever is higher.</p>



<h2 class="wp-block-heading">Damage and Liability for Compensation</h2>



<p>Even where the GDPR has been infringed, the injured party does not automatically acquire a right to compensation. Under Article 82 GDPR, three conditions must be met cumulatively: an infringement of the Regulation, material or non-material damage, and a causal link between the two. The controller is liable; the processor is liable only where it has failed to comply with obligations specifically directed to processors or has acted outside or contrary to the lawful instructions of the controller (Article 82(2)). Exemption from liability is possible only on proof that the entity concerned is not in any way responsible for the event giving rise to the damage (Article 82(3)). Where more than one controller or processor is responsible for the damage, they are jointly and severally liable (Article 82(4)).</p>



<p>The case law of the CJEU and of the Polish courts on Article 82 GDPR has so far developed mainly in cases concerning the unauthorised disclosure of data to third parties; there are no decisions dealing directly with the loss or corruption of data. The principles developed by the Court are, however, general in nature and apply to all types of breach listed in Article 4(12) GDPR, and thus also to breaches of the integrity and availability of data. The most important of them are:</p>



<ul class="wp-block-list">
<li>C-300/21 (Österreichische Post) &#8211; a mere infringement is not sufficient for an award of compensation; the claimant must show damage (material or non-material) and a causal link; at the same time, there is no minimum threshold of seriousness of the damage;</li>



<li>C-687/21 (BL v MediaMarktSaturn Hagen-Iserlohn) &#8211; the mere fear or apprehension of possible misuse of data by third parties may constitute non-material damage, but the claimant must prove that it actually occurred; such apprehension does not constitute damage where there was no possibility of a third party becoming aware of the data, which is typically the case where data has been destroyed or lost without being disclosed; compensation is purely compensatory and not punitive in nature;</li>



<li>judgment of the Warsaw Court of Appeal of 22 June 2023, case no. I ACa 352/23 &#8211; the court confirmed that non-material damage within the meaning of Article 82 GDPR also covers distress and the risk of reputational harm.</li>
</ul>



<p>The CJEU has also pointed out that Article 82 GDPR covers only infringements of the Regulation itself. If the loss or corruption of data additionally involves an infringement of national law, including of personality rights, any extension of liability can be considered only on the basis of national provisions; in Poland, the injured party may then claim separate compensation for non-material harm under Articles 23 and 24 in conjunction with Article 448 of the Civil Code.</p>



<h1 class="wp-block-heading">Liability for Loss or Corruption of Non-Personal Data</h1>



<h2 class="wp-block-heading">Contractual Liability</h2>



<p>The basis for claims against external providers whose improper performance has contributed to the loss or corruption of data is Article 471 of the Polish Civil Code. This applies in particular to software vendors, cloud service operators, IT outsourcing companies and entities responsible for backup and disaster recovery. Under that provision, the debtor is obliged to remedy the damage resulting from non-performance or improper performance of an obligation, unless it proves that this is the consequence of circumstances for which it is not responsible.</p>



<p>In commercial practice the key instrument governing the scope of that liability is the agreement concluded with the IT service provider (e.g. a maintenance or implementation agreement) and, within it, the Service Level Agreement (SLA). It is in the SLA that the parties define the scope of the service, the required availability of systems, the permitted response time in the event of a failure and the consequences of non-compliance, including contractual penalties and grounds for termination. From the perspective of liability for data loss, two SLA parameters are of particular importance:</p>



<ul class="wp-block-list">
<li>Recovery Point Objective (RPO) &#8211; the acceptable amount of data an organisation can afford to lose in the event of an incident; this parameter determines the required frequency of backups;</li>



<li>Recovery Time Objective (RTO) &#8211; the maximum acceptable time for restoring operational processes after a disruption.</li>
</ul>



<p>Both parameters form part of disaster recovery, i.e. the set of policies, procedures and tools designed to minimise the effects of unforeseen events such as cyberattacks, infrastructure failures or natural disasters. Disaster recovery comprises three layers:</p>



<ul class="wp-block-list">
<li>prevention (backups, protective systems, maintenance),</li>



<li>detection (monitoring, security audits),</li>



<li>remediation (data restoration, activation of the business continuity plan).</li>
</ul>



<p>In the case of cloud services, which consist in providing storage space, computing power or ready-made applications on infrastructure managed by the provider, liability for the loss or corruption of data rests as a rule with the operator that has taken control of the infrastructure. The customer should pay particular attention to data sovereignty clauses: providers subject to US law may be required to disclose data to law enforcement authorities under the CLOUD Act, even if the data is stored outside the territory of the United States.</p>



<h2 class="wp-block-heading">Contractual Penalties, Limitation of Liability and Force Majeure</h2>



<p>A breach of the agreement with the provider gives rise to contractual liability. Where the agreement provides for a contractual penalty, improper performance obliges the provider to pay it. Where the breach has caused damage and no contractual penalty has been stipulated, or the right to claim damages exceeding the penalty has not been excluded, the customer may claim damages under Article 471 of the Civil Code. In the cases described here, improper performance will most often consist in exceeding the RTO or failing to meet other SLA parameters.</p>



<p>The parties may contractually modify the scope of the provider&#8217;s liability, in particular by:</p>



<ul class="wp-block-list">
<li>limiting liability to actual loss, excluding lost profits,</li>



<li>introducing a monetary cap on damages,</li>



<li>excluding the statutory warranty for defects.</li>
</ul>



<p>In practice, customers often stipulate that the contractor is liable without limit for the most serious breaches of the agreement, which include precisely the loss of data, breach of confidentiality obligations and infringement of intellectual property rights. Force majeure may relieve the provider of liability, but this requires proof that the event was external, unforeseeable and unavoidable even with the use of all available means. In the case of known methods of cyberattack or foreseeable infrastructure failures, such a defence may be difficult to sustain, which is why the parties very often define in their agreements the catalogue of circumstances treated as force majeure.</p>



<p>The contractual liability regime described above applies to both personal and non-personal data, but its role differs fundamentally depending on the type of data. For personal data, liability under Article 471 of the Civil Code supplements the GDPR regime: the controller may bring recourse claims against a provider whose negligence contributed to a breach for which the controller has been held liable towards data subjects or the supervisory authority. For non-personal data, the agreement and the contractual liability arising from it are often the only available source of legal protection, given the absence of specific statutory rules dedicated to this type of data. This means that the SLA provisions, the precision of the RPO and RTO parameters, the scope of the liability clauses and the contractual penalties determine whether the provider will be held liable towards the customer.</p>



<h1 class="wp-block-heading">Other Bases of Liability</h1>



<h2 class="wp-block-heading">Criminal Liability</h2>



<p>Under Article 268a of the Polish Criminal Code, criminal liability is incurred by anyone who destroys, damages, deletes, alters or hinders access to computer data, or significantly disrupts or prevents the automatic processing, collection or transmission of such data. The offence is prosecuted on the motion of the injured party, meaning that the injured entity must file a request for prosecution. Criminal liability therefore extends, for example, to a person who deletes or corrupts data in a company&#8217;s computer system. The type of data (personal or non-personal) is irrelevant to the existence of the offence. Under Article 269 of the Criminal Code, where the data is of particular importance for national defence, transport safety, the functioning of government administration, another state authority or state institution, or local government, the offence carries a heavier penalty.</p>



<h2 class="wp-block-heading">Liability of Members of Company Bodies</h2>



<p>Liability of management board members may also arise under Article 293 of the Polish Commercial Companies Code (limited liability company) and Article 483 of that Code (joint-stock company) for damage caused to the company by an act or omission contrary to the law or the articles of association. Such damage may include an administrative fine imposed on the company as a result of a breach caused, for example, by a failure to implement appropriate safeguards or a data protection policy, or by a failure to carry out a DPIA, as well as the need to pay compensation to data subjects.</p>



<h2 class="wp-block-heading">Loss or Corruption of Data as a Cybersecurity Incident</h2>



<p>The loss or corruption of data in a business is a serious cybersecurity incident. In view of the reliance of the modern economy on data and the growth of digital threats, the EU legislator places great emphasis on protecting the availability and integrity of all of a company&#8217;s digital assets, attaching severe public-law liability to their loss. The legal framework is currently set by three key instruments: the DORA Regulation (dedicated to the financial sector), the NIS2 Directive and the Polish Act on the National Cybersecurity System (UKSC) implementing it. These rules impose stringent obligations on organisations in three main areas:</p>



<ul class="wp-block-list">
<li>ICT risk management &#8211; entities covered by the rules are required to implement security management systems, continuously assess risk and monitor their systems for threats. Under Article 18(1)(d) DORA, data loss is one of the direct criteria for classifying an incident as major. Under both DORA and the UKSC, incidents must be reported to the competent authorities and, in the case of more serious incidents, also to clients.</li>



<li>Digital resilience and business continuity &#8211; organisations must have genuine mechanisms ensuring continuity of operations in a crisis, including backup procedures, emergency data restoration and recovery plans (for both personal and non-personal data) and systematic testing of protective systems.</li>



<li>Oversight of third-party providers &#8211; the rules (DORA in particular) significantly restrict cooperation with IT service providers that do not meet security standards. Agreements between financial entities and third-party providers must, among other things, guarantee the right to terminate where weaknesses in data protection are identified and must contain provisions on that protection.</li>
</ul>



<p>Failures in these areas resulting in the loss or corruption of data are no longer merely an operational problem. The amended UKSC provides for severe financial penalties for non-compliance, ranging from PLN 20,000 up to EUR 10,000,000.</p>



<h1 class="wp-block-heading">Conclusion</h1>



<p>The loss or corruption of data is never a legally neutral event. The scope and nature of liability, however, depend above all on the type of data affected by the incident and on whether the entity responsible for the data has complied with its obligations:</p>



<ul class="wp-block-list">
<li>personal data &#8211; administrative liability arises from an infringement of the GDPR, while liability for compensation depends on the existence of damage and a causal link;</li>



<li>non-personal data &#8211; the burden of protection rests on civil law and on the contract; the precision of the SLA provisions, the RPO and RTO parameters and the liability clauses determines whether the business can successfully pursue its claims, and in the absence of dedicated statutory rules the contract often remains the only instrument of protection;</li>



<li>regardless of the type of data &#8211; the loss or corruption of data may give rise, in parallel, to the criminal liability of the perpetrator, the liability of management board members towards the company and, for entities subject to sectoral regulation, public-law sanctions under DORA and the UKSC.</li>
</ul>



<p>An appropriate response to the incident, including timely notification of the breach to the competent authorities, may significantly reduce the scope of that liability.</p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/liability-for-data-loss-and-corruption/">Liability for Loss and Corruption of Data</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/liability-for-data-loss-and-corruption/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>A Human in the Loop, Data under Control. The GDPR and the AI Act in Implementation Practice</title>
		<link>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ai-act-and-gdpr-ai-governance-compliance/</link>
					<comments>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ai-act-and-gdpr-ai-governance-compliance/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Mon, 05 Oct 2026 09:55:26 +0000</pubDate>
				<category><![CDATA[IT, NEW TECHNOLOGIES, MEDIA AND COMMUNICATION TECHNOLOGY LAW]]></category>
		<category><![CDATA[AI Act]]></category>
		<category><![CDATA[AI Compliance]]></category>
		<category><![CDATA[AI Governance]]></category>
		<category><![CDATA[AI Regulation]]></category>
		<category><![CDATA[algorithmic bias]]></category>
		<category><![CDATA[Artificial intelligence]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[Data Protection Law]]></category>
		<category><![CDATA[DPIA]]></category>
		<category><![CDATA[EU AI Act]]></category>
		<category><![CDATA[FRIA]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[high-risk AI]]></category>
		<category><![CDATA[Human Oversight]]></category>
		<category><![CDATA[large language models]]></category>
		<category><![CDATA[LLM]]></category>
		<category><![CDATA[Machine Learning]]></category>
		<category><![CDATA[personal data]]></category>
		<category><![CDATA[PRIVACY]]></category>
		<category><![CDATA[workplace AI]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8947</guid>

					<description><![CDATA[<p>Publication date: October 05, 2026 Implementing artificial intelligence in an organisation does not begin with choosing a tool, but with answering the question of who we are in the AI value chain and what we are responsible for. This article sets out the relationship between the GDPR and the AI Act: from the roles of [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ai-act-and-gdpr-ai-governance-compliance/">A Human in the Loop, Data under Control. The GDPR and the AI Act in Implementation Practice</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color"><strong>Publication date: October 05, 2026</strong></mark></p>



<p>Implementing artificial intelligence in an organisation does not begin with choosing a tool, but with answering the question of who we are in the AI value chain and what we are responsible for. This article sets out the relationship between the GDPR and the AI Act: from the roles of provider, deployer and controller, through algorithmic risks, the limits of lawfulness in the workplace and the rights of individuals, to the DPIA, the FRIA and the AI Governance model. It reflects the legal position following the entry into force of the Polish Act on Artificial Intelligence Systems and the postponement of the AI Act application dates by Regulation (EU) 2026/1744.</p>



<span id="more-8947"></span>



<h2 class="wp-block-heading"><strong>Introduction</strong></h2>



<p>Rapid technological progress and the widespread deployment of systems based on artificial intelligence are permanently reshaping the landscape of modern economic and social life. The use of machine learning algorithms and large language models in professional processes opens up unprecedented opportunities for optimisation, allowing the automation of tasks that previously required prolonged human effort. This technological efficiency, however, brings significant ethical and legal challenges and calls into question existing standards of privacy protection. In an era of such profound digital transformation, the key task is to develop a regulatory framework that balances the drive for innovation against the imperative of protecting fundamental rights. It must be remembered that AI systems, despite their apparent infallibility, operate on data that may be erroneous, biased or out of date. As a result, algorithm-based decision-making processes directly affect the legal and personal situation of natural persons, which creates a need to define precisely who is responsible for the outcome of the technology.</p>



<p>Regardless of how advanced AI tools become, the paradigm of human oversight must remain the fundamental principle of safe implementation. A human being cannot be reduced to a mere recipient of output data; they must perform an active supervisory function and act as a guarantor of reliability and safety when the algorithm fails. The AI Act<a href="#_ftn1" id="_ftnref1">[1]</a> establishes rigorous procedures and imposes specific restrictions on providers, deployers and other links in the artificial intelligence value chain. In relation to the GDPR<a href="#_ftn2" id="_ftnref2">[2]</a>, the AI Act is complementary: it does not repeal existing data protection principles but refines them in a technological context. The AI Act adds an additional layer of product-related and ethical obligations, ensuring that algorithms are used with respect for the rights and freedoms of the individual. Much of this framework already applies: the prohibitions in Article 5 AI Act have applied since 2 February 2025 and the transparency obligations in Article 50 AI Act since 2 August 2026, while Regulation (EU) 2026/1744 (the so-called Digital Omnibus on AI) has postponed the application of the obligations for stand-alone high-risk systems listed in Annex III to 2 December 2027. In Poland, the national framework is set out in the Act of 3 July 2026 on Artificial Intelligence Systems (Journal of Laws, item 1003), which established the Commission for the Development and Safety of Artificial Intelligence (KRiBSI) as the market surveillance authority.</p>



<h2 class="wp-block-heading"><strong>1st         Who Is Who in the AI Ecosystem: Provider, Deployer, Controller</strong></h2>



<h3 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>a)    Provider or Deployer? The Answer Determines the Scope of Liability</strong></h3>



<p>Defining an organisation&#8217;s role in the artificial intelligence ecosystem is of key importance for determining its legal obligations precisely. It must be established whether a given entity acts as the provider of an AI system or is merely a deployer of a ready-made technological solution. Under Article 3(3) AI Act, a provider is a natural or legal person, public authority, agency or other body that develops an AI system (or a general-purpose AI model) or has it developed, and places it on the market or puts it into service under its own name or trademark, whether for payment or free of charge. The provider&#8217;s principal task is therefore to create the technology and ensure its compliance with rigorous safety and data quality requirements before the commercialisation stage. Most organisations using AI systems will not act as providers but as deployers. This distinction is fundamental to risk analysis: while the provider is responsible for the architecture and the training process (compliance of the “product”), the deployer is responsible for how the system is operated, for instructions to employees and for the lawfulness of the data entered into the model during the inference phase.</p>



<p>Under Article 3(4) AI Act, a deployer may be a natural or legal person, public authority, agency or other body that uses an AI system under its authority. These features clearly distinguish the provider of an AI system from an entity that merely uses the supplied system and exercises authority over it. In addition, an entity using artificial intelligence systems may acquire the status of a controller of personal data once it meets the definitional criteria set out in Article 4(7) GDPR. A controller is therefore a natural or legal person, public authority or other body which, alone or jointly with others, determines the purposes and means of the processing of personal data. It should be stressed that an AI system itself has no legal personality, and responsibility for its actions rests with the entity that decides on its deployment and operating parameters. Data processing in this context takes place through operations or sets of operations performed on personal data by automated means, which falls within the broad definition in Article 4(2) GDPR. These processes include, in particular, the collection, recording, adaptation, consultation and use of data by algorithms in order to generate results (<em>output</em>). In the case of AI systems, this processing is highly automated and is usually aimed at optimising and increasing the operational efficiency of information processes, which places a particular duty on the controller to keep control over every stage of the data life cycle.</p>



<h3 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>b)    When an AI System Becomes a High-Risk System</strong></h3>



<p>Under the system adopted in the AI Act, the correct classification of artificial intelligence systems is of key importance for organisations, since solutions classified as high-risk systems are subject to the most stringent legal regime. An AI system acquires that status if it meets the conditions set out in Article 6(1) AI Act (this concerns systems that are products, or safety components of products, subject to EU certification) or if it is expressly listed in Annex III to the Regulation. That catalogue covers areas of critical importance for fundamental rights, such as biometrics and the categorisation of persons, the management of critical infrastructure, and education and vocational training. Particular attention should be paid to employment and workers management, access to essential public and commercial services, and systems used in law enforcement, migration management and the administration of justice. Classifying a technology in this category obliges the controller not only to ensure high-quality data and human oversight in accordance with the AI Act, but also to comply strictly with the GDPR, which in most cases will include carrying out a full data protection impact assessment (DPIA).</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>c)     Transparency, Instructions for Use and Human Oversight: Obligations in Practice</strong></h2>



<p>Providers must ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the persons concerned are informed that they are interacting with an AI system (Article 50(1) AI Act). High-risk AI systems must also be designed and developed in accordance with the principle of transparency of their operation (Article 13 AI Act), which enables deployers to interpret the system&#8217;s output and use it appropriately. It is also essential to attach instructions for use in an appropriate format, containing complete and comprehensible information for deployers. While AI systems are in use, it must be possible for natural persons to oversee the system (Article 14 AI Act) in order to prevent risks to health, safety or fundamental rights.</p>



<p>Deployers of high-risk AI systems must implement appropriate technical and organisational measures to ensure that they use such systems in accordance with the instructions for use supplied by the provider (Article 26(1) AI Act). Under Article 26(11) AI Act, deployers of the high-risk AI systems referred to in Annex III that make decisions, or assist in making decisions, related to natural persons must fulfil their information obligation before actually using the high-risk AI system, by clearly informing those persons that such technology is being used in relation to them.</p>



<p>In order to give full effect to the requirements of the AI Act, deployers are also obliged to assign human oversight to natural persons who have the necessary competence, training and authority, as well as the necessary organisational support. Under Article 14(2) of the Regulation, “<em>human oversight shall aim to prevent or minimise the risks to health, safety or fundamental rights (…)</em>”. Article 14(4), in turn, requires providers to deliver a high-risk AI system to the deployer in such a way as to enable the natural persons to whom human oversight is assigned, among other things, “<em>to remain aware of the possible tendency of automatically relying or over-relying on the output produced by a high-risk AI system (automation bias), in particular for high-risk AI systems used to provide information or recommendations for decisions to be taken by natural persons</em>”.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>d)    Training and Inference: Where the Data Risk Hides in Language Models</strong></h2>



<p>The foundation of large language models (LLMs) is the processing of vast quantities of data, enabling the generation of text and the performance of advanced natural-language tasks. As tools belonging to generative artificial intelligence, these models operate on the basis of the statistical prediction of successive tokens (words or parts of words) in response to instructions entered by the user, known as prompts. In market and corporate practice, the most common are the GPT series models (OpenAI), implemented in solutions such as ChatGPT and Microsoft Copilot, as well as the Llama family of models (Meta), DeepSeek and the Polish projects Bielik and PLLuM. From a data protection perspective, the key distinction is between the two fundamental phases of a model&#8217;s operation: training and inference.</p>



<p><strong>Training phase</strong></p>



<p>In this phase, the substantive “intelligence” of the model is created. The process is based on the analysis of enormous data sets (Big Data) in order to detect statistical correlations between tokens. It should be emphasised that an LLM does not interpret content cognitively (in the human sense) but calculates the probability of the next element in a sequence. The data is often obtained through the mass collection of content from the open internet (web scraping). From the GDPR perspective, this is the most controversial stage, because the controller of personal data (usually the provider of the system) must demonstrate a specific legal basis and purpose for the processing (Article 6 GDPR). At this stage the provider becomes responsible for the quality of the data sets. Under Article 10 AI Act, training data sets must be relevant, sufficiently representative and, to the best extent possible, free of errors and complete in view of the intended purpose of the system.</p>



<p><strong>Inference phase</strong></p>



<p>The inference phase is the operational stage in which the trained model generates answers to specific user queries (prompts), using the statistical weights developed during training. In this phase the model does not, as a rule, “learn” in real time but processes the information supplied to it at a given moment. From the perspective of an organisation deploying an AI system, this is the area of the highest legal risk. The key problem is employees entering customer data, personal information or trade secrets into queries sent to an external model. Such conduct constitutes a transfer of data to the provider&#8217;s servers, which often involves a transfer outside the European Economic Area (EEA). In this scenario the organisation acquires the status of a deployer within the meaning of Article 3(4) AI Act. Under Article 4(7) GDPR, the organisation, as the controller of its employees&#8217; and customers&#8217; data, bears full responsibility for the content of the information that reaches the AI system. So while the training phase remains the domain of large technology corporations (providers), the inference phase is the area in which every business bears direct responsibility for the security and lawfulness of the prompts it generates.</p>



<h2 class="wp-block-heading"><strong>2nd    Bias, Hidden Scoring and Explainability: Algorithmic Risks under Article 5 GDPR and Article 10 AI Act</strong></h2>



<p>The practical implementation of artificial intelligence systems in socially sensitive areas, such as the recruitment of employees or the assessment of creditworthiness, requires the systematic management of the risk of algorithmic error. The key challenge is to ensure the explainability of AI and to combat effectively the phenomenon of bias, which can lead to the discrimination of entire social groups. The foundation for counteracting these risks under data protection law is the set of principles expressed in Article 5(1) GDPR. Under the principle of transparency (point (a)), the organisation must be able to explain to the data subject the logic behind an autonomous decision of the algorithm. The principle of data minimisation (point (c)), in turn, requires the information collected to be limited to what is necessary for the purpose, which prevents models from being fed with redundant data or data that has no objective bearing on the result of the analysis. In this context, the prohibition of so-called hidden scoring takes on particular significance; this means an AI system covertly assessing, for example, job candidates by assigning them points without ensuring adequate transparency of the process. The controller, as the employer, is obliged under Article 13(2)(f) GDPR to provide the data subject with information about the existence of automated decision-making, including profiling (referred to in Article 22(1) and (4)), and meaningful information about the logic involved, as well as the significance and the envisaged consequences of such processing for the data subject.</p>



<p>AI systems frequently rely on profiling mechanisms, that is, the automated processing of personal data consisting of the use of that data to evaluate certain aspects of a natural person, in particular their competence, expected work performance or other characteristics relevant to the purpose for which the AI tool is used. Under Article 22(1) GDPR, “<em>the data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her</em>”, unless the decision is necessary for entering into or performing a contract, is authorised by Union or Member State law, or is based on the data subject&#8217;s explicit consent (Article 22(2) GDPR). Profiling as such is not prohibited in principle, but its use is subject to significant restrictions, particularly where it leads to decisions with significant effects on the person being profiled. In practice, this means that special care must be taken in designing and using AI tools, particularly with regard to the risk of bias. These systems learn from historical data, which may lead to the replication of earlier decision patterns and to unintended discrimination, for example on grounds of age, state of health or other personal characteristics. It is therefore essential to ensure transparency of the systems&#8217; operation, adequate human oversight and the ability to verify and challenge the results generated by AI.</p>



<p>Relying solely on the rules contained in the GDPR is not enough, however, when confronted with deep neural networks, which is why a key role is played by Article 10 AI Act, which governs data and data governance in high-risk AI systems. That provision imposes rigorous obligations on providers to examine and verify training, validation and testing data. These data sets must be relevant, sufficiently representative and, to the best extent possible, free of errors and complete in view of the specific geographical, behavioural or functional setting in which the system is to operate. This requirement directly compels organisations to apply procedures for detecting and neutralising hidden bias (bias mitigation). For example, a recruitment algorithm trained on the historical data of a company in which mainly men were promoted may classify the female gender as a negative feature; Article 10 AI Act requires such anomalies to be actively monitored and corrected at the design stage and through continuous oversight of the system&#8217;s operation.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>3rd      Can an Employer Require the Use of AI? The Limits of Lawfulness in the Workplace</strong></h2>



<p>Assessing the lawfulness of deploying artificial intelligence systems in the workplace requires a precise delineation of the boundary between the employer&#8217;s managerial prerogatives and the fundamental rights of employees. Whether an employee can be required to use AI tools depends on the nature of the tool and the extent of its intrusion into the employee&#8217;s privacy. While issuing a work instruction to use AI as simple support in daily duties falls within the law, the deployment of monitoring, evaluation or biometric systems encounters strict legal barriers. The legal basis for deploying standard algorithms that optimise work processes is most often Article 6(1)(f) GDPR, that is, the legitimate interest of the controller (the employer). Its application is conditional, however, on passing the so-called balancing test, which must show that the employer&#8217;s economic or organisational interests do not override the autonomy, rights and freedoms of the persons employed. In the case of high-risk AI tools, the balancing test will generally be harder to pass, and legitimate interest alone may prove insufficient, particularly where the processing leads to decisions of the kind referred to in Article 22 GDPR.</p>



<p>The situation becomes radically more complicated where an AI system requires the processing of biometric data in order to operate (for example, facial recognition systems used to record working time or to analyse an employee&#8217;s emotions). Biometric data belong to the special categories of data whose processing is, as a rule, prohibited under Article 9(1) GDPR. For an employer to use such data lawfully, one of the exceptional conditions in Article 9(2) GDPR must be met; in the reality of Polish labour law this most often requires the employee&#8217;s explicit and fully voluntary consent (subject to the requirements of Article 22¹ᵇ of the Labour Code) or a specific legal provision imposing such an obligation on safety grounds. The employee&#8217;s lack of genuine freedom of choice in the relationship with the employer makes such consent extremely difficult to defend before the supervisory authorities.</p>



<p>The ultimate barrier to the implementation of high-risk algorithmic AI systems is Article 5 AI Act, which introduces a categorical list of practices prohibited in the European Union. On that basis it is unlawful to deploy AI systems used for so-called <em>social scoring</em> (the point-based classification of citizens or employees on the basis of their social behaviour), as well as systems that carry out biometric categorisation in order to infer sensitive characteristics (for example political opinions or sexual orientation). Moreover, in the context of employment relationships, the AI Act prohibits the use of emotion recognition systems in the workplace (Article 5(1)(f) AI Act), except for systems put in place for medical or safety reasons, which in practice closes the door to employers algorithmically testing the mood or stress levels of their staff. These prohibitions have applied since 2 February 2025, and Regulation (EU) 2026/1744 extends their catalogue from 2 December 2026.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>4th       Individuals versus the Algorithm: Erasure, Human Intervention, Transparency</strong></h2>



<p>The informational and technological asymmetry between the individual and the entities deploying artificial intelligence systems requires effective instruments of legal protection. The greatest controversy in this area currently surrounds the exercise of rights of an erasure nature. Enforcing the right to be forgotten under Article 17 GDPR encounters a fundamental technical barrier in the form of the so-called “memory” of generative models and neural networks. In traditional databases, deleting information means erasing a specific record. In the case of AI models, input data becomes irreversibly integrated into the structure of the model&#8217;s mathematical weights during the training process. Reversing that state, described in the literature as machine unlearning, is a complex and costly process and often simply impossible without retraining the model in its entirety. This creates a deep conflict between the individual&#8217;s right to demand the erasure of their data and the architecture of modern technologies.</p>



<p>Another key safeguard is Article 22 GDPR, which lays down a general prohibition on subjecting natural persons to decisions based solely on automated processing, including profiling, which produce legal effects concerning them or similarly significantly affect them. This provision gives the individual the right to obtain human intervention (<em>human-in-the-loop</em>), to express their own point of view and to contest a decision taken by an algorithm (for example, an automated refusal of credit or rejection of a job application). Supervisory authorities&#8217; guidelines and the case law of the CJEU (including the judgment of 7 December 2023 in Case C-634/21 SCHUFA Holding) indicate that human involvement in the decision-making process may not be fictitious or token; a manager or analyst must have a real ability to change the verdict generated by the AI. These guarantees are complemented, at the level of direct interaction with the technology, by the information obligations arising from Article 50 AI Act, applicable since 2 August 2026. That provision imposes strict operational transparency requirements on providers and deployers of artificial intelligence systems. Under it, every natural person interacting with an AI system, such as a chatbot or voice assistant, must be clearly and promptly informed of that fact, unless this is obvious from the context of use. Equally strict obligations apply to the labelling of synthetically generated content, including so-called deepfakes. Deployers of AI systems that generate or manipulate images, audio or video content in such a way that it closely resembles authentic persons or events are legally required to disclose that the content has been artificially generated. The purpose of this regulation is to protect the cognitive autonomy of the individual and to counter mass disinformation.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>5th       Four Risk Tiers and a New Supervisory Map: KRiBSI alongside the President of the Personal Data Protection Office</strong></h2>



<p>A key element of the EU strategy for regulating artificial intelligence is the risk-based approach, which directly determines the scope of the legal obligations imposed on organisations. The Artificial Intelligence Act introduces a rigid division of systems into four categories: prohibited practices (posing an unacceptable risk to fundamental rights), high-risk systems, limited-risk systems (mainly subject to information obligations) and minimal-risk systems. From the operational perspective of businesses and public institutions, the most important category is that of high-risk systems, whose definition and classification mechanism are governed by Article 6 AI Act. That provision refers directly to Annex III to the AI Act, which contains a closed catalogue of critical areas, including workers management and access to employment (HR), education and vocational training, as well as law enforcement and the management of critical infrastructure. Implementing AI tools in these sectors triggers a strict legal regime, including the need to implement a risk management system, ensure high-quality data and provide human oversight. Under Regulation (EU) 2026/1744, these obligations will apply to stand-alone systems listed in Annex III from 2 December 2027 and to systems embedded in products covered by Annex I from 2 August 2028; the postponement does not, however, change the scope of the requirements but merely gives organisations time to implement them.</p>



<p>From the perspective of mapping risk within an organisation, this classification has profound consequences under data protection law. The Act of 3 July 2026 on Artificial Intelligence Systems entrusted market surveillance to the Commission for the Development and Safety of Artificial Intelligence (KRiBSI), which from 28 October 2026 acquires full powers of inspection and the power to impose the fines provided for in the AI Act. This does not, however, exclude the President of the Personal Data Protection Office (PUODO). Under Article 74(8) AI Act, in respect of the high-risk AI systems listed in Annex III, point 1 (biometrics, in so far as used for law enforcement, migration and the administration of justice), point 6 (law enforcement), point 7 (migration and asylum) and point 8 (administration of justice and democratic processes), market surveillance is linked to the data protection authorities, and the PUODO consistently emphasised, already in its comments on the draft act, that any processing of personal data in AI systems remains within its competence under the GDPR. This means that a compliance audit cannot be limited to a standard data protection impact assessment (DPIA). The organisation must take account of a dual supervision model in which KRiBSI verifies the system&#8217;s compliance with the AI Act, while the data protection authority retains full investigative and corrective powers in respect of the processing of personal data, including the data used to build the model. Ignoring this dual competence at the system design stage (<em>privacy and compliance by design</em>) exposes controllers to a serious risk of financial penalties and administrative sanctions in the form of an order to shut down the algorithm immediately.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>6th       DPIA and FRIA: from a Document in a Drawer to a Dialogue with the Regulator</strong></h2>



<p>Because of their systemic and often unpredictable impact on privacy, the deployment of advanced artificial intelligence systems in most cases automatically triggers an obligation to carry out a data protection impact assessment. Under Article 35 GDPR, that assessment (Data Protection Impact Assessment, DPIA) is mandatory whenever processing using new technologies is likely to result in a high risk to the rights or freedoms of natural persons, which, in the context of recruitment or credit-assessment algorithms (mentioned above in relation to the principles of minimisation and transparency in Article 5 GDPR), is an operational standard. The contemporary AI legal regime, however, dramatically extends that obligation. Under Article 27 AI Act, public bodies and private entities providing public services, as well as deployers of the high-risk AI systems referred to in Annex III, point 5(b) and (c), are required to carry out a rigorous fundamental rights impact assessment (FRIA). The entities referred to in Annex III, point 5(b) and (c) AI Act are those using high-risk AI systems to evaluate the creditworthiness of natural persons or establish their credit score, and systems used for risk assessment and pricing in relation to natural persons in the case of life and health insurance. This mechanism cannot operate in isolation from the GDPR; in practice, the FRIA becomes an integral, deeper extension of the classic DPIA, requiring the organisation to examine the impact of the algorithm not only on privacy but also on human dignity, non-discrimination and the right to a fair trial, that is, the impact on the fundamental rights of the persons in relation to whom the AI systems are used.</p>



<p>The traditional approach, in which the DPIA was merely an internal, static corporate document filed away in a drawer, is becoming a thing of the past when confronted with Article 57(10) AI Act, which governs so-called regulatory sandboxes. These sandboxes, which are controlled testing environments for innovative AI solutions, redefine the relationship between the controller and the supervisory authority. An organisation that decides to participate in a regulatory sandbox is legally obliged to cooperate closely and transparently with the authority operating the sandbox (in Poland, KRiBSI) and, in so far as the system under test processes personal data, also with the President of the Personal Data Protection Office (Article 57(10) AI Act). In this new model of cooperation, the DPIA process evolves into a tool for a dynamic, multilateral dialogue with the state authority. The results of the risk analysis are consulted on an ongoing basis, which makes it possible to eliminate algorithmic bias under the regulator&#8217;s eye before the system is commercialised. Such a compliance model not only minimises the risk of severe financial penalties but also makes it possible to build AI systems that are safe and transparent from the stage of their technological incubation.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>7th       AI Governance: How to Implement AI in an Organisation and Be Able to Prove It</strong></h2>



<p>Effective management of artificial intelligence systems within an organisation requires a multi-level oversight model (AI Governance<a href="#_ftn3" id="_ftnref3">[3]</a>) that combines technical protective measures with a systematic assessment of legal risk. The foundation for designing safe algorithms is formed by the principles of data protection by design and by default, codified in Article 25 GDPR. They oblige the organisation to take privacy protection and data minimisation into account at the earliest stage of creating the architecture of the AI model, and also when selecting training data and model parameters. Article 32 GDPR, in turn, which lays down the principle of security of processing, takes on an entirely new technological dimension in the world of autonomous systems. Classic IT safeguards (such as encryption or access control) are no longer sufficient against the specific vulnerabilities of neural networks. Ensuring the integrity and confidentiality of data requires controllers to actively harden models against a new generation of cyberattacks, including <em>adversarial attacks</em><a href="#_ftn4" id="_ftnref4">[4]</a>, attempts to poison training data sets (<em>data poisoning</em>) and model inversion techniques aimed at extracting the personal data used to build the model.</p>



<p>The formal tool for verifying these safeguards and estimating the risks to natural persons remains the data protection impact assessment (DPIA) carried out under Article 35 GDPR. Because of the informational asymmetry and intrusiveness of AI models, however, this assessment ceases to be merely an internal, static document and becomes a platform for continuous monitoring. For artificial intelligence systems classified as high-risk solutions, the traditional DPIA must be integrated with the new, strict obligation under Article 27 AI Act: the fundamental rights impact assessment (FRIA). Only the synergy of the technical resilience of the model with a transparently conducted DPIA/FRIA procedure enables an organisation to demonstrate compliance and lawfully keep an AI system in commercial use.</p>



<h1 class="wp-block-heading"><strong>Summary and Conclusions</strong></h1>



<p>The analysis shows that the implementation of artificial intelligence systems, including advanced large language models (LLMs), redefines existing standards of privacy protection and requires organisations to build entirely new governance structures. The key to ensuring the lawfulness of data processing is the precise identification of the entity&#8217;s role in the AI ecosystem. The distinction between the provider, responsible for the training phase and the quality of the product data, and the deployer, which as controller bears full responsibility for the inference phase and the content of the prompts entered, is the starting point for effectively mapping regulatory risk under the GDPR and the Artificial Intelligence Act.</p>



<p>The main challenge for modern organisations is the management of specific algorithmic risks, such as the lack of explainability of machine decisions and hidden bias, which in sensitive areas such as recruitment or credit assessment may lead to the replication of discriminatory social patterns. Traditional data protection principles, in particular the principles of transparency and minimisation, gain strong support from Article 10 AI Act. That provision revolutionises compliance processes by imposing strict requirements for the verification of training data, with the aim of actively neutralising bias at the system design stage. Furthermore, the right to be forgotten (Article 17 GDPR) encounters a technological barrier in the form of the mathematical memory of neural networks, which means that the exercise of individuals&#8217; rights against the algorithm requires advanced procedures such as machine unlearning, and the unconditional guarantee of real rather than merely token human oversight (<em>human-in-the-loop</em>).</p>



<p>Equally important is the categorisation of systems by level of risk. Classifying AI tools as high-risk (for example in the employment sector or financial scoring) triggers strict information obligations and absolute prohibitions of impermissible practices, which include emotion recognition systems in the workplace and biometric profiling to infer sensitive characteristics. In this context, the classic data protection impact assessment (DPIA) is evolving: for high-risk systems it must be integrated with the fundamental rights impact assessment (FRIA) under Article 27 AI Act. In addition, the analysis of institutional roles points to a model of supervision shared between the newly established Commission for the Development and Safety of Artificial Intelligence (KRiBSI) and the President of the Personal Data Protection Office (PUODO), which retains full competence wherever an AI system processes personal data, while risk assessment processes cease to be static documents and become part of a dynamic dialogue with the regulator, including within regulatory sandboxes.</p>



<p>Consequently, the lawful and ethical use of artificial intelligence in modern business is not possible without a comprehensive oversight model: <em>AI Governance</em>. Traditional IT security measures must be extended to include mechanisms that harden models against a new generation of cyberattacks, such as <em>adversarial</em> attacks and <em>data poisoning</em>. The synergy of the principles of <em>privacy by design</em> and <em>privacy by default</em> with the rigorous procedures of the AI Act and the GDPR leads to the conclusion that the EU legal framework does not block innovation but civilises the process of digital transformation. It guarantees that technological development takes place within the limits of the law, with respect for the personhood and cognitive autonomy of the human being.</p>



<h2 class="wp-block-heading"><strong>Bibliography</strong></h2>



<h2 class="wp-block-heading"><strong>Legislation</strong></h2>



<ol class="wp-block-list">
<li>Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (OJ L 119, 4.5.2016, p. 1, as amended) – hereinafter: GDPR</li>



<li>Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA relevance) (OJ L, 2024/1689, 12.7.2024) – hereinafter: AI Act</li>



<li>Act of 26 June 1974 – Labour Code (consolidated text: Journal of Laws of 2025, item 277, as amended).</li>
</ol>



<h2 class="wp-block-heading"><strong>Online Sources</strong></h2>



<ul class="wp-block-list">
<li>Letter of [date] issued by the President of the Personal Data Protection Office, ref. DOL.401.354.2024, <em>Comments of the PUODO on the draft Act on Artificial Intelligence Systems</em>, www.uodo.gov.pl.</li>



<li>Wolters Kluwer: “<em>Przetwarzanie danych osobowych na podstawie prawnie uzasadnionego interesu administratora</em>” [Processing of personal data on the basis of the controller&#8217;s legitimate interest], 12.12.2025, available online: https://www.lex.pl/test-rownowagi-rodo,43864.html</li>



<li>European Parliamentary Research Service (EPRS): <em>Artificial Intelligence Act: State of play and implementation challenges, European Parliamentary Research Service (EPRS) At a Glance Briefing</em>, Brussels 2026, [online], available at: https://www.europarl.europa.eu/RegData/etudes/ATAG/2026/785673/EPRS_ATA(2026)785673_EN.pdf (accessed: 29.05.2026).</li>



<li>Biznes Myśli: <em>Fine-tuning LLM – fakty i mity. Jak skutecznie dotrenować duży model językowy</em> [Fine-tuning LLMs – facts and myths. How to fine-tune a large language model effectively], [online], available at: https://biznesmysli.pl/fine-tuning-llm-fakty-i-mity/ (accessed: 29.05.2026).</li>



<li>LegalGeek: <em>AI Act vs RODO. Jak sztuczna inteligencja wpływa na ochronę danych osobowych</em> [AI Act vs GDPR. How artificial intelligence affects personal data protection], Legal Knowledge Portal for Business, [online], available at: https://legalgeek.pl/blog/ai-act-vs-rodo/ (accessed: 29.05.2026).</li>



<li>Wikipedia (the free encyclopedia): <em>Duży model językowy</em> [Large language model], [online], available at: https://pl.wikipedia.org/wiki/Du%C5%BCy_model_j%C4%99zykowy (accessed: 29.05.2026).</li>



<li>Data Science Robię: <em>AI Governance – co to takiego i dlaczego jest teraz kluczowe?</em> [AI Governance – what is it and why is it now key?], [online], available at: https://www.datasciencerobie.pl/ai-governance-co-to-takiego/ (accessed: 02.06.2026).</li>
</ul>



<hr class="wp-block-separator has-alpha-channel-opacity"/>



<p><a href="#_ftnref1" id="_ftn1">[1]</a> Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (Text with EEA relevance) (OJ L, 2024/1689, 12.7.2024) – hereinafter: AI Act</p>



<p><a href="#_ftnref2" id="_ftn2">[2]</a> Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (OJ L 119, 4.5.2016, p. 1, as amended) – hereinafter: GDPR</p>



<p><a href="#_ftnref3" id="_ftn3">[3]</a> Data Science Robię: AI Governance – co to takiego i dlaczego jest teraz kluczowe? [AI Governance – what is it and why is it now key?], [online], available at: https://www.datasciencerobie.pl/ai-governance-co-to-takiego/ (accessed: 02.06.2026).</p>



<p><a id="_ftn4" href="#_ftnref4">[4]</a> adversarial attacks – the deliberate manipulation of input data in order to deceive the algorithm.</p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ai-act-and-gdpr-ai-governance-compliance/">A Human in the Loop, Data under Control. The GDPR and the AI Act in Implementation Practice</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/ai-act-and-gdpr-ai-governance-compliance/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
		<item>
		<title>The MyDr Data Breach – Lessons for Healthcare Service Providers and e-Health Platforms</title>
		<link>https://www.kg-legal.eu/info/pharmaceutical-healthcare-life-sciences-law/mydr-data-breach-lessons-for-healthcare-providers-kglegal/</link>
					<comments>https://www.kg-legal.eu/info/pharmaceutical-healthcare-life-sciences-law/mydr-data-breach-lessons-for-healthcare-providers-kglegal/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Mon, 07 Sep 2026 20:30:11 +0000</pubDate>
				<category><![CDATA[PHARMACEUTICAL, HEALTHCARE & LIFE SCIENCES LAW]]></category>
		<category><![CDATA[Act on Patients Rights]]></category>
		<category><![CDATA[AI Act]]></category>
		<category><![CDATA[AI cybersecurity]]></category>
		<category><![CDATA[AI in healthcare]]></category>
		<category><![CDATA[AI-enabled attacks]]></category>
		<category><![CDATA[anomaly detection]]></category>
		<category><![CDATA[anonymisation]]></category>
		<category><![CDATA[anonymization]]></category>
		<category><![CDATA[Artificial intelligence]]></category>
		<category><![CDATA[breach notification]]></category>
		<category><![CDATA[cloud healthcare]]></category>
		<category><![CDATA[controller liability]]></category>
		<category><![CDATA[Cyber Five]]></category>
		<category><![CDATA[cyber insurance]]></category>
		<category><![CDATA[Cyber Resilience]]></category>
		<category><![CDATA[cyber resilience healthcare]]></category>
		<category><![CDATA[cyber risk]]></category>
		<category><![CDATA[Cyber Security]]></category>
		<category><![CDATA[cybersecurity certification]]></category>
		<category><![CDATA[cybersecurity insurance]]></category>
		<category><![CDATA[data breach]]></category>
		<category><![CDATA[data breach response]]></category>
		<category><![CDATA[data controller]]></category>
		<category><![CDATA[data minimisation]]></category>
		<category><![CDATA[data minimization]]></category>
		<category><![CDATA[Data Privacy]]></category>
		<category><![CDATA[data processing agreement]]></category>
		<category><![CDATA[data processor]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[data protection impact assessment]]></category>
		<category><![CDATA[deepfake]]></category>
		<category><![CDATA[digital health]]></category>
		<category><![CDATA[digital healthcare]]></category>
		<category><![CDATA[DPA]]></category>
		<category><![CDATA[DPIA]]></category>
		<category><![CDATA[e-health]]></category>
		<category><![CDATA[eHealth platforms]]></category>
		<category><![CDATA[electronic medical records]]></category>
		<category><![CDATA[EMR security]]></category>
		<category><![CDATA[EU AI Act]]></category>
		<category><![CDATA[gdpr]]></category>
		<category><![CDATA[GDPR Article 28]]></category>
		<category><![CDATA[GDPR Article 32]]></category>
		<category><![CDATA[GDPR Article 33]]></category>
		<category><![CDATA[GDPR Article 34]]></category>
		<category><![CDATA[GDPR Article 35]]></category>
		<category><![CDATA[GDPR Article 82]]></category>
		<category><![CDATA[GDPR Compliance]]></category>
		<category><![CDATA[GDPR compliance healthcare]]></category>
		<category><![CDATA[health data]]></category>
		<category><![CDATA[health tech]]></category>
		<category><![CDATA[Healthcare AI]]></category>
		<category><![CDATA[healthcare compliance]]></category>
		<category><![CDATA[healthcare contracts]]></category>
		<category><![CDATA[healthcare cybersecurity]]></category>
		<category><![CDATA[healthcare data breach]]></category>
		<category><![CDATA[healthcare data security]]></category>
		<category><![CDATA[healthcare IT]]></category>
		<category><![CDATA[healthcare providers]]></category>
		<category><![CDATA[healthcare software]]></category>
		<category><![CDATA[healthcare technology]]></category>
		<category><![CDATA[identity verification]]></category>
		<category><![CDATA[Incident Response]]></category>
		<category><![CDATA[Information Security]]></category>
		<category><![CDATA[information security management]]></category>
		<category><![CDATA[ISO 27001]]></category>
		<category><![CDATA[IT supplier contracts]]></category>
		<category><![CDATA[joint and several liability]]></category>
		<category><![CDATA[liability insurance]]></category>
		<category><![CDATA[medical data protection]]></category>
		<category><![CDATA[medical records]]></category>
		<category><![CDATA[medical technology]]></category>
		<category><![CDATA[MedTech]]></category>
		<category><![CDATA[MFA]]></category>
		<category><![CDATA[multi-factor authentication]]></category>
		<category><![CDATA[MyDr data breach]]></category>
		<category><![CDATA[National Cybersecurity System]]></category>
		<category><![CDATA[network segmentation]]></category>
		<category><![CDATA[NIS2]]></category>
		<category><![CDATA[patient data]]></category>
		<category><![CDATA[patient privacy]]></category>
		<category><![CDATA[patients rights]]></category>
		<category><![CDATA[Personal data breach]]></category>
		<category><![CDATA[personal data protection]]></category>
		<category><![CDATA[phishing]]></category>
		<category><![CDATA[phishing attacks]]></category>
		<category><![CDATA[Poland cybersecurity]]></category>
		<category><![CDATA[Polish GDPR]]></category>
		<category><![CDATA[Polish healthcare law]]></category>
		<category><![CDATA[privacy by design]]></category>
		<category><![CDATA[processor liability]]></category>
		<category><![CDATA[pseudonymisation]]></category>
		<category><![CDATA[pseudonymization]]></category>
		<category><![CDATA[Regulatory Compliance]]></category>
		<category><![CDATA[risk assessment]]></category>
		<category><![CDATA[Risk Management]]></category>
		<category><![CDATA[sensitive personal data]]></category>
		<category><![CDATA[special category data]]></category>
		<category><![CDATA[sub-processors]]></category>
		<category><![CDATA[subprocessor management]]></category>
		<category><![CDATA[supplier risk management]]></category>
		<category><![CDATA[technology contracts]]></category>
		<category><![CDATA[telemedicine]]></category>
		<category><![CDATA[third-party liability]]></category>
		<category><![CDATA[UODO]]></category>
		<category><![CDATA[vendor risk]]></category>
		<category><![CDATA[voice deepfake]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8909</guid>

					<description><![CDATA[<p>Publication date: September 07, 2026 What happened On 10 August 2026, MyDr, one of Poland&#8217;s largest providers of electronic medical records software, confirmed that it had been the target of a deliberate criminal attack on its systems. Two days later, the Ministry of Digital Affairs announced that the incident may affect close to 19 million [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/pharmaceutical-healthcare-life-sciences-law/mydr-data-breach-lessons-for-healthcare-providers-kglegal/">The MyDr Data Breach – Lessons for Healthcare Service Providers and e-Health Platforms</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: September 07, 2026</mark></strong></p>



<h2 class="wp-block-heading"><strong>What happened</strong></h2>



<div class="wp-block-media-text is-stacked-on-mobile" style="grid-template-columns:42% auto"><figure class="wp-block-media-text__media"><video controls src="https://www.kg-legal.eu/wp-content/uploads/2026/09/generated-video-1-1.mp4"></video></figure><div class="wp-block-media-text__content">
<p>On 10 August 2026, MyDr, one of Poland&#8217;s largest providers of electronic medical records software, confirmed that it had been the target of a deliberate criminal attack on its systems. Two days later, the Ministry of Digital Affairs announced that the incident may affect close to 19 million individuals and more than 12,000 healthcare facilities, and that the exfiltrated database exceeds 2 terabytes.</p>
</div></div>



<p>The attackers supplied the security portal Zaufana Trzecia Strona with a data sample suggesting that they hold PESEL numbers (Polish national identification numbers) and at least fragments of prescription information.</p>



<span id="more-8909"></span>



<p>To date, the company has not publicly confirmed the full scope and nature of the compromised data, referring instead to a pending forensic analysis. The investigation is being conducted by the Central Bureau for Combating Cybercrime under the supervision of the Warsaw Regional Prosecutor&#8217;s Office, and the President of the Personal Data Protection Office (UODO) has opened an inspection covering the technical and organisational measures applied and the underlying risk analysis. Since 29 August, the dataset from the incident has been available on the government portal bezpiecznedane.gov.pl, where anyone can check whether their data was affected.</p>



<p>The scale of the incident prompted the Ministry of Digital Affairs to announce, within three weeks, a legislative package branded the &#8220;Cyber Five&#8221;. Its key elements include certification of entities processing medical data within the existing national cybersecurity certification framework; a mandatory risk assessment before processing begins and at least every two years thereafter; new obligations for entities serving more than 100 controllers or processing data of more than 100,000 individuals (including rapid transfer of affected persons&#8217; data to CSIRT NASK and a duty to inform client facilities about the level of their own security); and notifications of medical events via the mObywatel and mojeIKP applications. The amendments are to cover the Act on Patients&#8217; Rights and the legislation governing the National Cybersecurity System.</p>



<h2 class="wp-block-heading"><strong>Why roles in the processing chain decide everything</strong></h2>



<p>From a legal standpoint, the critical point is that, in relation to medical records, MyDr acts as a processor, while each facility – from a large clinic network to a single-doctor practice – remains the controller. The consequences of this structure became fully apparent after the incident:</p>



<ul class="wp-block-list">
<li><strong>The obligation to notify UODO (Article 33 GDPR) and to communicate the breach to patients (Article 34 GDPR)</strong> rests with the controller, i.e. the facility. The processor is merely required to inform the controller &#8220;without undue delay&#8221; (Article 33(2) GDPR). In practice, thousands of healthcare providers had to assess risk and communicate with patients on the basis of fragmentary information from the supplier.</li>



<li><strong>Liability for damages (Article 82 GDPR)</strong> is joint and several: a patient may sue the facility, the supplier, or both. The processor is liable where it has failed to comply with obligations specifically imposed on processors by the GDPR or acted contrary to the controller&#8217;s instructions – but the facility is liable for having entrusted data to an entity that did not provide sufficient guarantees (Article 28(1) GDPR).</li>



<li><strong>Sector-specific requirements under the Act on Patients&#8217; Rights</strong> (Article 24(4)) impose on any entity entrusted with the processing of medical records a duty of confidentiality and data security – this is precisely the provision now slated for extension to include certification and minimum technological standards.</li>
</ul>



<p>Deputy Minister of Digital Affairs Dariusz Standerski stated openly that in this case, liability under the contract remained entirely with the controllers, i.e. small medical practices. This is the most important lesson of the incident: a data processing agreement is not a formality but the document that, on the day of a breach, determines who pays.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>How a healthcare service provider can protect itself</strong></h2>



<p><strong>1. The data processing agreement and the main contract as risk-allocation tools.</strong> Standard DPA templates offered by software vendors focus on satisfying the minimum requirements of Article 28(3) GDPR. A healthcare provider should negotiate further: a precise deadline and format for incident notification (e.g. 24 hours, with a defined scope of information enabling a risk assessment); audit and penetration-testing rights; an obligation to maintain specified certifications (ISO 27001 and, in future, certification under the National Cybersecurity System Act); a duty to cooperate in communications with patients and the supervisory authority; liability and recourse clauses not capped at the annual fee; and a requirement that the vendor hold cyber insurance with a defined sum insured, with the facility named as a co-insured or beneficiary.</p>



<p><strong>2. Cyber and liability insurance.</strong> Standard professional liability policies for healthcare providers typically do not cover the cost of notifying patients, crisis management, administrative fines or claims arising from data breaches. A dedicated cyber policy covers these elements, but its exclusions must be read carefully: insurers increasingly condition cover on the implementation of MFA, system patching and backups, and an incident at an external supplier (a so-called <em>third-party breach</em>) is often covered only under an express extension. It is also worth verifying whether the software vendor&#8217;s own policy actually exists and what its limit is – given the number of facilities relying on a single system, such amounts may prove illusory.</p>



<p><strong>3. A map of relationships between entities.</strong> In a real-world e-health ecosystem, patient data flows between the facility, the EMR vendor, the hosting or cloud provider, the e-prescription and e-referral operator, laboratories, IT subcontractors and billing companies. Each link is a distinct legal relationship: processing on behalf of the controller, sub-processing (Article 28(2) and (4) GDPR) or joint controllership (Article 26 GDPR). A facility should maintain an up-to-date register of these entities, know where the data is physically located and control the chain of sub-processors – a &#8220;general&#8221; consent to sub-processors without a list and without a right to object is, in practice, an abdication of control.</p>



<p><strong>4. Anonymisation, pseudonymisation and data minimisation.</strong> Data that is not in the system cannot leak. Healthcare providers and vendors should separate identifiers (PESEL numbers, contact details) from clinical data, apply pseudonymisation (Article 4(5) and Article 32(1)(a) GDPR) in test, analytical and research environments, and store statistical data exclusively in anonymised form. It should be remembered that anonymisation is an irreversible process and only such a process removes data from the scope of the GDPR; pseudonymisation remains processing of personal data, but it significantly limits the consequences of a breach and is a valuable argument both in proceedings before UODO and in litigation over damages.</p>



<p><strong>5. Internal obligations and incident readiness.</strong> A breach response procedure should be tested, not merely written down: who decides on notifying UODO within 72 hours, who communicates with patients, who with the media, who secures the evidence. Regular risk analysis and a data protection impact assessment (DPIA) for EMR systems – which, given their scale and the categories of data involved, almost always meet the criteria of Article 35 GDPR – is an obligation already today, and once the &#8220;Cyber Five&#8221; enters into force it will additionally become a sector-specific requirement with a prescribed frequency.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>AI-enabled attacks</strong></h2>



<p>The healthcare sector is particularly exposed to a new generation of AI-supported attacks. Large language models enable the mass generation of credible phishing messages in flawless Polish, personalised on the basis of data from previous breaches – a PESEL number, a surname and information about a prescription are enough to construct a convincing message &#8220;from your clinic&#8221; or &#8220;from the National Health Fund&#8221;. AI tools also automate the discovery of vulnerabilities in systems and the generation of malicious code, shortening the window between disclosure of a vulnerability and its exploitation. There is a growing number of cases involving voice deepfakes used to impersonate medical staff or IT administrators in order to obtain access credentials.</p>



<p>For vendors and facilities, this means that traditional &#8220;don&#8217;t click suspicious links&#8221; training is no longer sufficient. Technical mechanisms are required (phishing-resistant MFA, network segmentation, AI-assisted anomaly monitoring on the defensive side), together with identity verification procedures for every request for data access or a change of permissions. Regulatory risk should also be kept in mind: AI systems deployed in healthcare facilities – including tools supporting diagnostics or triage – fall under the AI Act, and their integration with EMR systems constitutes yet another link in the processing chain that must be reflected in contracts and in the risk analysis.</p>



<h2 class="wp-block-heading has-luminous-vivid-amber-background-color has-background"><strong>How we support institutional clients</strong></h2>



<p>KG Legal advises healthcare providers, medical networks, telemedicine platform and e-health software vendors, and investors in this sector on managing liability for data. Our support includes auditing existing data processing agreements and IT supplier contracts for risk allocation; negotiating liability, recourse and insurance clauses; mapping the chain of processors and vetting subcontractors; preparing and testing breach response procedures; handling notifications to UODO and communications with patients; and representation in inspection proceedings and in damages litigation. For medical technology vendors, we prepare documentation and contract templates meeting the requirements of the GDPR, the Act on Patients&#8217; Rights, NIS2 and – once enacted – the &#8220;Cyber Five&#8221; provisions, and we assess the compliance of AI-based solutions with the AI Act and the MDR. Our aim is that, on the day an incident occurs, the client knows exactly who is responsible for what and has evidence of having exercised due diligence.</p>



<p><em>Facts as at 2 September 2026, based on statements by MyDr, the Ministry of Digital Affairs and UODO, and press reports. This article is for information purposes only and does not constitute legal advice.</em></p>
<p>#MyDr #DataBreach #HealthcareCybersecurity #HealthcareData #DataProtection #GDPR #Cybersecurity #HealthTech #eHealth #DigitalHealth #MedicalRecords #EMR #PatientData #PatientPrivacy #UODO #Poland #CyberRisk #CyberInsurance #IncidentResponse #DataPrivacy #NIS2 #CyberFive #ISO27001 #AIAct #AIinHealthcare #AICybersecurity #MedTech #Telemedicine #CyberResilience #DataSecurity</p>
<p> </p>
<p> </p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/pharmaceutical-healthcare-life-sciences-law/mydr-data-breach-lessons-for-healthcare-providers-kglegal/">The MyDr Data Breach – Lessons for Healthcare Service Providers and e-Health Platforms</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/pharmaceutical-healthcare-life-sciences-law/mydr-data-breach-lessons-for-healthcare-providers-kglegal/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		<enclosure url="https://www.kg-legal.eu/wp-content/uploads/2026/09/generated-video-1-1.mp4" length="492961" type="video/mp4" />

			</item>
		<item>
		<title>Our Report on the Legal Landscape of the Drone Industry Is Today’s Main Feature on Lexology</title>
		<link>https://www.kg-legal.eu/info/kg-legal-news/our-report-on-the-legal-landscape-of-the-drone-industry-is-todays-main-feature-on-lexology/</link>
					<comments>https://www.kg-legal.eu/info/kg-legal-news/our-report-on-the-legal-landscape-of-the-drone-industry-is-todays-main-feature-on-lexology/#respond</comments>
		
		<dc:creator><![CDATA[jakub]]></dc:creator>
		<pubDate>Mon, 07 Sep 2026 12:02:57 +0000</pubDate>
				<category><![CDATA[KG LEGAL NEWS]]></category>
		<category><![CDATA[AI Act]]></category>
		<category><![CDATA[Artificial intelligence]]></category>
		<category><![CDATA[Aviation Law]]></category>
		<category><![CDATA[data protection]]></category>
		<category><![CDATA[defence financing]]></category>
		<category><![CDATA[Defence Industry]]></category>
		<category><![CDATA[drone industry]]></category>
		<category><![CDATA[Drone Law]]></category>
		<category><![CDATA[Drone Regulation]]></category>
		<category><![CDATA[Dual-Use Technology]]></category>
		<category><![CDATA[export controls]]></category>
		<category><![CDATA[legal landscape]]></category>
		<category><![CDATA[Lexology]]></category>
		<category><![CDATA[Regulatory Compliance]]></category>
		<category><![CDATA[regulatory landscape]]></category>
		<category><![CDATA[SANCTIONS]]></category>
		<category><![CDATA[Technology Law]]></category>
		<category><![CDATA[UAS regulation]]></category>
		<category><![CDATA[UAV Law]]></category>
		<guid isPermaLink="false">https://www.kg-legal.eu/?p=8904</guid>

					<description><![CDATA[<p>Publication date: September 07, 2026 We are pleased to share that our in-depth report on the legal and technological landscape of the drone industry is featured today as the leading content on Lexology&#8217;s homepage. The report examines the rapidly evolving drone ecosystem through nine interconnected legal and regulatory pillars, including aviation law, artificial intelligence, cybersecurity, [&#8230;]</p>
<p>Artykuł <a href="https://www.kg-legal.eu/info/kg-legal-news/our-report-on-the-legal-landscape-of-the-drone-industry-is-todays-main-feature-on-lexology/">Our Report on the Legal Landscape of the Drone Industry Is Today’s Main Feature on Lexology</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></description>
										<content:encoded><![CDATA[
<p><strong><mark style="background-color:rgba(0, 0, 0, 0)" class="has-inline-color has-vivid-cyan-blue-color">Publication date: September 07, 2026</mark></strong></p>



<figure class="wp-block-image size-large"><img decoding="async" width="1024" height="602" src="https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY-1024x602.png" alt="" class="wp-image-8905" srcset="https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY-1024x602.png 1024w, https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY-300x176.png 300w, https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY-768x451.png 768w, https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY-1536x902.png 1536w, https://www.kg-legal.eu/wp-content/uploads/2026/09/AI-DRONE-WARFARE-LEXOLOGY.png 1583w" sizes="(max-width: 1024px) 100vw, 1024px" /></figure>



<span id="more-8904"></span>



<p>We are pleased to share that our in-depth report on the legal and technological landscape of the drone industry is featured today as the leading content on Lexology&#8217;s homepage.</p>



<p>The report examines the rapidly evolving drone ecosystem through nine interconnected legal and regulatory pillars, including aviation law, artificial intelligence, cybersecurity, data protection, sanctions, export controls, dual-use technologies and defence-sector financing.</p>



<p>As drone technologies advance faster than many existing regulatory frameworks, companies, investors and advisers must navigate an increasingly complex environment spanning civilian, dual-use and military applications.</p>



<p>We invite you to read the full report on our blog, available without a paywall:</p>



<p><blockquote class="wp-embedded-content" data-secret="V2PD940UnY"><a href="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/drone-warfare/">Drone Warfare</a></blockquote><iframe class="wp-embedded-content" sandbox="allow-scripts" security="restricted"  title="&#8222;Drone Warfare&#8221; &#8212; KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019" src="https://www.kg-legal.eu/info/it-new-technologies-media-and-communication-technology-law/drone-warfare/embed/#?secret=LBenNwlEFP#?secret=V2PD940UnY" data-secret="V2PD940UnY" width="500" height="282" frameborder="0" marginwidth="0" marginheight="0" scrolling="no"></iframe> </p>



<p>#DroneLaw #DroneTechnology #UAS #ArtificialIntelligence #Cybersecurity #DataProtection #DualUse #DefenceIndustry #AviationLaw #ExportControls #Sanctions #RegulatoryCompliance #TechnologyLaw #Innovation #LegalAnalysis</p>
<p> </p>




<p>Artykuł <a href="https://www.kg-legal.eu/info/kg-legal-news/our-report-on-the-legal-landscape-of-the-drone-industry-is-todays-main-feature-on-lexology/">Our Report on the Legal Landscape of the Drone Industry Is Today’s Main Feature on Lexology</a> pochodzi z serwisu <a href="https://www.kg-legal.eu">KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019</a>.</p>
]]></content:encoded>
					
					<wfw:commentRss>https://www.kg-legal.eu/info/kg-legal-news/our-report-on-the-legal-landscape-of-the-drone-industry-is-todays-main-feature-on-lexology/feed/</wfw:commentRss>
			<slash:comments>0</slash:comments>
		
		
			</item>
	</channel>
</rss>
