Karolina Bednarska

Introduction – The Reform of the Polish National Labour Inspectorate: Key Changes in 2026

New provisions governing the operation of the Polish National Labour Inspectorate (Państwowa Inspekcja Pracy, hereinafter: “PIP”) have been in force since 8 July 2026. The Ministry of Family, Labour and Social Policy (hereinafter: “MRPiPS”) is changing the law in order to strengthen the protection of employee rights in Poland. The amendment does not alter the Labour Code itself; rather, it introduces tools for the effective and efficient enforcement of the law already in force. Minister Agnieszka Dziemianowicz-Bąk described the act as a further instrument protecting honest businesses against unfair competitors who for years have built their advantage at the expense of workers – preventing the artificial lowering of labour costs (social security contributions, taxes, paid leave). The changes also implement milestones of Poland’s National Recovery Plan (KPO) aimed at curbing the abuse of civil-law contracts and improving the protection of persons in employment. The Ministry’s measures are directed above all at limiting bogus self-employment and so-called “junk contracts”.

Following the entire legislative process – on the website of the Government Legislation Centre – one can see that the Ministry’s bill received mixed reviews; opinions were divided. The Association of Labour Inspectors of the Republic of Poland (hereinafter: “SIPRP”)[1] assessed the bill amending the Act on the National Labour Inspectorate rather sceptically, with criticism clearly predominating. Although the Association had for years called for a deep reform of the PIP, in its view the act in its present shape still requires improvement. The SIPRP considers that the new duties imposed on the Inspectorate are not matched by adequate tools to strengthen its position. It also emphasised repeatedly in its opinion that certain provisions are causing serious concern within the profession (for example, the modification of the rules on the conclusion of inspection proceedings – Article 11(7a) of the PIP Act), and that the bill lacked the solutions for which the profession had been waiting for years and for which it had appealed in vain (in the form of petitions and systemic postulates aimed at genuinely strengthening the position of the PIP). A similar tone marked the opinion prepared by the Trade Unions Forum (hereinafter: “FZZ”),[2] which stressed the lack of dialogue between the government side and civil society organisations. The solutions adopted are, in its view, deprived of the voice of the people who apply the provisions drawn up by the Ministry. Nevertheless, the FZZ notes certain solutions positively (for example, the shift from a model of immediate administrative authority to a two-stage procedure with an order to remedy infringements, and the transfer of decision-making competence to the higher, district level), while drawing attention to practical and logistical issues, such as a realistic estimate of costs, the need for staffing support (new posts at the PIP) and improvements in digitalisation and civil procedure, so that the courts do not become the “bottleneck” of the reform. The Polish Economic Society – Employers (hereinafter: “PTG”)[3] also spoke out on the reform of the National Labour Inspectorate. In its assessment, the government’s reforming measures create a risk of infringing economic freedom and freedom of contract. Part of the business community fears that officials will gain excessive discretionary power over what is, and what is not, a contract of mandate. Many entrepreneurs are concerned that the new provisions will strike at B2B arrangements which operate in accordance with the law, are fully lawful, and are chosen deliberately by specialists.

The opinions of these various organisations show clearly that the amendment of the provisions on the National Labour Inspectorate is not a reform to which the market and businesses can be indifferent. It represents a long-awaited chance to curb abuses and to improve the situation of persons engaged on fictitious civil-law contracts. Time will tell whether the new provisions prove an effective tool in the fight against abuses or – as business fears – a source of costly disputes and legal uncertainty.

Contracts under PIP scrutiny – what other powers do inspectors gain? What New Powers Does the PIP Have?

The most important reform of the PIP’s powers is the conferral on district labour inspectors (okręgowi inspektorzy pracy) of the power to issue decisions converting improperly concluded contracts into employment contracts. The inspectors’ decision will be preceded by a multi-stage process. First, where infringements are detected, the PIP issues an order to remedy the irregularities. The inspector sets an appropriate time limit for concluding an employment contract, or for arranging the actual terms of cooperation in such a way that they comply with the law. Importantly, following the change in the law the burden of proof has been reversed: it will not be the inspector who has to demonstrate that the arrangement is in fact an employment relationship; rather, the employer will have to defend itself against a finding that an employment relationship exists and prove that the relationship in question is genuinely an independent B2B arrangement.

If the irregularities are not remedied by the business, the PIP inspector applies to the District Labour Inspector for a decision converting the improperly concluded contract into an employment contract. Both the employer and the worker will be able to appeal against the decision to the labour court, which will stay its enforcement until judgment (unless the decision has been declared immediately enforceable). For the duration of the court proceedings, the amendment provides for the possibility of securing the claim: the contract between the parties may be varied, terminated by notice or dissolved only in accordance with the rules of labour law. Until the dispute is finally resolved, a person covered by such an interim measure cannot be deprived of protection merely because he or she formally concluded a civil-law contract. In the event of notice of termination or dissolution of the contract, that person is entitled to the judicial remedies provided for in labour law. These measures are intended to protect workers’ rights, since an employment contract guarantees, among other things: paid annual leave; paid maternity, paternity and parental leave; paid sick leave; and protection against unjustified dismissal and against termination of the contract from one day to the next. In practice this may materially change the total cost of engagement – overtime pay, and the calculation of social security (ZUS) contributions and personal income tax advances.

The PIP may inspect any company, enterprise or firm, regardless of its size or industry. Its inspections will not focus solely on an analysis of the shape of the contract – its name and its wording. The inspectors’ activities are to concentrate on what the cooperation between the “employer” and the “worker” looks like in practice. The key elements will be subordination, the place and time of performance of the work, the personal provision of services, the continuity of the cooperation, the manner of supervision, and who actually bears the economic risk and organises the work. Scrutiny in this area will be facilitated by the fact that, following the changes in the law, the PIP is able to carry out inspections remotely and without prior notice, drawing on information from the ZUS, CEIDG, KAS and KRUS systems. In practice, this synergy of supervisory authorities will mean that data on employment and business relationships will be cross-checked, and irregularities detected automatically. It is therefore worth putting documents and procedures in order before any inspection begins. Even if an employer amends the provisions of a contract in order to shield itself from reclassification as an employment contract, there is no guarantee that the PIP will not nevertheless change the contract by decision after analysing the particular case. If the inspectors find that the day-to-day working model resembles regular employment and that the relevant contractual clauses do not alter the facts on the ground, the National Labour Inspectorate will have grounds to convert such a contract.

The reform introduced by the MRPiPS is not intended to abolish either contracts of mandate or B2B contracts. Concluding such contracts remains lawful and they may still be used, but they must correspond to the real model of cooperation. The PIP’s new powers are meant to counteract pathologies in the labour market in which a civil-law contract is used as a form of engagement in place of an employment contract, even though the manner in which the work is performed points unmistakably to an employment relationship.

The act also introduces solutions such as an individual interpretation (advance ruling) procedure. An employer will be able to apply – the interpretations are issued by the Chief Labour Inspector – for confirmation that the way in which it engages its staff complies with the law. If irregularities come to light in the course of the proceedings, the authority will indicate how to bring the situation into conformity with the law. The interpretation is issued by way of a decision, against which an appeal lies to the court on the terms laid down in the Code of Civil Procedure. It is binding in principle, but it does not preclude an inspection if the facts established in its course differ from those described in the application. Compliance with the interpretation provides protection against administrative and financial sanctions and penalties within the scope covered by the interpretation. In addition, since the entry into force of the amendment the PIP has been able to conduct remote inspections, which is intended to streamline and speed up inspections; and the maximum fine that the PIP may impose in summary penal-ticket proceedings has been at least doubled – from PLN 2,000 to PLN 5,000, and up to PLN 10,000 in cases of repeat offending. The grant of these broad competences was accompanied by the imposition on the institution of a significant duty: the PIP must draw up annual and multi-annual action plans for targeted inspections on the basis of risk analysis.

The impact of the new regulations on businesses operating in Poland

The PIP reforms have a significant impact on business operations, including the activity of foreign entities employing or posting workers in Poland. The new inspection environment compels a far-reaching modification of organisational structures, compliance procedures, and the approach to legal risk management.

International capital groups and their subsidiaries should impose on the entities acting on their behalf in Poland corporate procedures governing the establishment of cooperation and the drafting of contracts with Polish personnel. A foreign company can no longer rely solely on the autonomy of local management without oversight. What is required is the establishment of a permanent audit trail, the appointment of dedicated HR managers, and the alignment of corporate policy with the realities of Polish legislation. Foreign enterprises must move away from flexible but risky models of cooperation based exclusively on civil-law (B2B) contracts for tasks of a continuous nature performed under supervision. This requires converting such relationships into standard employment wherever the hallmarks of an employment relationship are present.

Foreign employers who post their workers to Poland may be inspected by labour inspectors both as to the lawfulness of the posting (documentation, notifications in the EU systems) and as to whether a genuine act of posting actually took place (verification that the posting is not fictitious). Owing to the large number of reports concerning sham agencies, the PIP is also focusing its activity on the verification of such institutions; it is therefore important to maintain the consistency, continuity and credibility of documentation. A lack of consistency in documentation carries the risk of the workers being deemed to be employed directly in Poland, with all the financial consequences that entails.

Foreign entities engaging persons who perform remote work from Polish territory must take account of the question of the permanent place of performance of their worker’s duties, with a view to the correct settlement of taxes and social insurance. The cooperation agreement should set out the terms of engagement in detail, because the PIP – in addition to analysing employment contracts and their practical application – verifies whether remote work is not concealing a permanent branch of the foreign enterprise in Poland, a so-called permanent establishment for tax purposes, which is decisive for determining where the tax obligation lies.

Temporary employment agencies may also find themselves under the particular supervision of the National Labour Inspectorate. Concluding civil-law contracts for work of a continuous nature performed under the supervision of the user undertaking is contrary to the law and subject to rectification. Offering temporary work on the basis of civil-law contracts instead of temporary employment contracts exposes firms to immediate decisions changing the nature of the engagement and to the withdrawal of the agency’s operating licence.

Firms which use the services of contractors (independent providers rendering services under a B2B agreement or contract) or of outsourcing (companies to which selected tasks are contracted out) must pay close attention to clauses indicative of employee subordination. If a contractor performs repetitive tasks at a designated place and time under the direction of the principal, the contract may be challenged as a circumvention of labour law.

It becomes crucial to introduce detailed, consistent and transparent documentation: personnel files, working time records, documents confirming the actual performance of duties, and internal decision-making paths. Furthermore, international capital groups must implement advanced internal compliance procedures which align corporate policy with Polish labour law – something that requires constant monitoring of the conformity of HR practices with the legislation.

Changes in the law need not automatically mean risk, provided that the company prepares for them early and comprehensively. The key to certainty and peace of mind is, above all, preventive action and diligent documentation. To protect an enterprise against HR-related problems, the first step is to carry out a regular audit of all civil-law contracts and existing practices; this allows potential irregularities to be detected relatively quickly. The analysis covers an assessment of clauses concerning subordination, the place and time of performance of tasks, reporting rules, and the scope of the provider’s independence, making it possible to eliminate potential risk factors before an inspector takes up the matter. Once the audit has been completed, new contract templates should be implemented: a safe contract should clearly specify the scope of liability, the option of using subcontractors, the absence of organisational subordination, and the bearing of economic risk by the provider. Equally important is the gathering of evidence of the entire course of the cooperation between the parties (preferably in digital form). In the event of an inspection, it is precisely the organisational documentation that will be decisive.

If, however, the scope of duties and the manner of performing tasks point unequivocally to an employment relationship, the company must take the strategic decision to convert the defective contracts into employment contracts. Taking that decision sufficiently early may protect the firm from having to pay arrears of contributions with interest and from the imposition of sanctions. An alternative tool for contracts giving rise to doubt is to use the individual interpretation mechanism referred to in Article 14b of the amended PIP Act.

Regardless of the type of contract chosen, the inspectors’ attention may be drawn by a complaint lodged by a former associate after the relationship with the firm has ended. The longer the cooperation lasted, and the more it resembled regular employment, the higher the risk that the documentation presented will not withstand scrutiny by the authority. Serious doubts on the part of the PIP, ZUS and the tax office are also raised by the mass transfer of salaried employees onto B2B contracts or contracts of mandate, which tends to be interpreted as a circumvention of the law aimed at avoiding public-law burdens. It is also worth checking whether there is any divergence in the company’s records between the wording of the contract and internal documents: if the contract guarantees full freedom, but company regulations oblige associates to take part in training, annual appraisals or the use of the social benefits fund, the inspector will analyse these materials together – and internal reality will always prevail over what is written on paper.

Whatever the reason for the commencement of inspection activities by the National Labour Inspectorate, what is key for the business, and for the proper course of the visit, is to adopt an open and matter-of-fact attitude and to provide the inspector with appropriate conditions for the efficient verification of documents and the examination of witnesses. It should be remembered, however, that constructive cooperation does not mean the uncritical acceptance of every demand made. The business has every right to expect that all activities will be conducted strictly within the limits of the inspector’s authorisation, in accordance with the applicable procedure, and with respect for all the legal guarantees to which the inspected entity is entitled.

Inspection of the legality of employment

The new provisions focus on the elimination of the so-called grey zone and of abuses at the interface between civil law and labour law. The inspectors’ examination of the legal relationship linking the “employer” and the “worker” for lawfulness is carried out on the basis of the criteria in Article 22 of the Labour Code: what is decisive is solely the sum of the characteristics of the actual performance of the tasks, not the name of the contract or its formal provisions. An employment relationship rests on subordination (the giving of instructions as to the manner, place and time of performance of the work), the personal performance of work (no possibility of appointing a substitute without the employer’s consent), continuity and remuneration (pay disbursed at fixed intervals irrespective of the result or effect of the work), and the employer’s responsibility (provision of the tools of work, infrastructure, and so on). Faced with such rigorous criteria for the assessment of contracts, businesses are under urgent pressure to review their existing models of cooperation. Ignoring these regulations may lead to serious financial and legal consequences, which is why the implementation of proactive adjustment measures becomes crucial.

B2B contracts are currently the most frequently challenged form of cooperation. The PIP pays particular attention to elements such as:

• exclusivity – the service provider renders services to one entity only, which accounts for 100% of its revenue;
• fixed hours and compulsory presence at the premises of the ordering party, principal or client, with no possibility of remote work or a flexible schedule, even though the contract provides otherwise;
• incorporation into the company’s organisational structure – a specific assigned position, a business e-mail box on the company’s domain, conducting conversations with clients in the company’s name without disclosing the status of external provider, and so on;
• the ordering firm, principal or client supplies equipment, tools and company resources (e.g. a laptop, a business telephone, access to systems, software and infrastructure) on the same terms as to salaried employees;
• absence of any real economic risk – the provider issues an invoice for the same amount every month, regardless of the number of hours and the scope of the services performed; there are no contractual penalties; it bears no business costs apart from ZUS contributions; it invests no capital of its own;
• ongoing direction – the issuing of work instructions which cannot be refused.

The contract for a specific work (umowa o dzieło), in turn – even though it is a distinctive form of cooperation, since what counts is a specific and visible result – remains within the National Labour Inspectorate’s field of interest. Inspectors pay attention to elements such as:

• the result of the contract – if the “work” consists in the monthly provision of services, ongoing support and the like, a contract for a specific work will not be the correct form of cooperation; the work must be capable of being identified, verified and accepted on delivery;
• repetitiveness and continuity of performance – a “work” delivered every month, of the same value and of a similar scope, means that in practice we are dealing with a contract of mandate or an employment relationship;
• the work lacks the features of individual creativity, being a copy or a multiplication of something;
• there is constant supervision over the manner in which the work is created – the provider reports progress daily, takes part in briefings and receives instructions.

Following the amendment of the PIP Act, the contract most exposed to challenge is the contract of mandate (umowa zlecenia), since of all civil-law contracts its construction is the closest to an employment relationship. The difference lies in the degree of subordination and independence – and that is precisely what the inspector examines. The most frequently challenged elements are:

• a fixed rota and compulsory attendance;
• an obligation to perform the mandate in person, with no possibility of subcontracting;
• hourly settlement with a working time register (where the records serve not to settle the service but to control working time, this signals the possible existence of employer direction);
• an obligation to justify absences and to apply for leave (a break in the provision of services may be contractual, but its organisation should not replicate statutory annual leave);
• a fixed monthly remuneration independent of the scope of performance (regardless of the number of hours, the tasks performed or the results achieved);
• ongoing work instructions and daily supervision.

The key to avoiding adverse legal and financial consequences lies not in superficial touch-ups to the wording of contracts, but in a genuine reordering of day-to-day practices and in guaranteeing the actual independence of providers wherever civil-law contracts are used.

EU labour law and the PIP reform

The reform of the National Labour Inspectorate in Poland is closely linked to the harmonisation of Polish law with EU regulations. The Directive on improving working conditions in platform work (2024/2831) introduces a legal presumption of an employment relationship for persons working through online platforms. If the application controls and directs the work, the person is deemed to be an employee rather than self-employed – and the burden of proof rests on the platform. Nowadays this form of work is increasingly chosen by young people, which compels both the EU and the Polish legislator to take protective measures in this field. Once the directive is transposed into Polish law (the transposition deadline is 2 December 2026), the PIP will have a tool for enforcing this presumption in practice. The directive additionally prohibits the collection of certain personal data (for example, on emotional state) by algorithms and requires human oversight of decisions to block or suspend accounts. The introduction of these rules in the platform work directive forms part of the broader European digital package and is closely connected with the provisions of the Digital Services Act (DSA). While the directive protects labour rights and prevents algorithmic arbitrariness towards service providers, the DSA imposes on digital platforms general requirements of transparency, accountability and oversight of algorithms, which complement the protection of users and digital workers. As a result, the two acts together form a coherent system: the DSA looks after a safe and transparent digital environment at the level of the platform as a whole, while the directive translates those principles directly into work relationships and fair conditions of engagement.

Also important in the context of the PIP’s activity is Directive 2019/1152 on transparent and predictable working conditions. It introduced an information obligation (Article 29 of the Labour Code) under which, within 7 days of the employee being admitted to work, the employer must provide the employee with detailed information on the terms of employment. In this context labour inspectors may verify whether that information is complete and exhaustive – for example, whether it includes information on overtime and the compensation for it, the rules for moving from shift to shift (in the case of shift work), information on the right to training and on the company’s training policy, or the rules for determining the length of the probationary period and daily and weekly rest, and so on. Moreover, under this directive the inspectors have obtained tools to check whether the termination of an employment contract was not the consequence of the employee asserting rights due under the Labour Code or of lodging a complaint with the PIP; they may also examine clauses in employment contracts and workplace regulations for prohibited bans on working for other entities, review the correctness of the conclusion of probationary contracts, and check whether the employer reimburses the costs of, or correctly accounts for the working time spent on, compulsory training.

In the context of the posting of workers, the EU legislator has adopted two significant directives: Directive 96/71/EC and Directive 2018/957/EU amending it. One effect of the work on these directives, and of the accompanying EU enforcement framework, is an easier exchange of data between the PIP and the authorities of other Member States; moreover, it is on the basis of this framework that ZUS facilitates the verification of the pay and working conditions of cross-border workers. A directive of key importance, however, is the one on the protection of whistleblowers, which has been a direct impulse for the initiation of targeted inspections by the PIP. Directive (EU) 2019/1937 protects whistleblowers – persons who report breaches of the law – ensuring their safety from retaliation and guaranteeing the confidentiality of their identity. It guarantees protection against dismissal, demotion or other retaliatory measures on the part of the employer, as well as the confidentiality of the identity of the reporting person and of persons connected with them.

In the context of the EU directives mentioned above, an important European Union agency is the European Labour Authority (hereinafter: “ELA”), which forms a bridge connecting the National Labour Inspectorate with the inspectorates of other Member States. The Authority supports the PIP above all by organising and coordinating joint and concerted cross-border inspections, and by streamlining the exchange of information in cases of abuse concerning the posting of workers or bogus self-employment. The ELA’s activity also means that the Polish inspectorate gains real operational backing for the enforcement of the EU standards mentioned above in the international dimension.

The EU regulations presented here form a coherent, modern system of labour protection, with the result that the Polish labour inspectorate obtains better tools and its role correspondingly gains in importance. The PIP is ceasing to be merely an authority reacting to traditional shortcomings in production plants, and is becoming a modern guardian of the law, equipped with the tools to combat digital exploitation, opaque algorithms and cross-border abuses. The reform of the National Labour Inspectorate is thus a natural and necessary response to the challenges of today’s dynamically changing labour market.

Conclusion

The reform of the powers of the National Labour Inspectorate is one of the most important turning points in contemporary Polish employment policy, carrying with it both great hopes and considerable controversy. On the one hand, the new legal instruments – such as the possibility of converting civil-law contracts into employment by administrative decision – are meant genuinely to curb the abuse of “junk contracts” and to protect workers better. On the other hand, employers fear excessive bureaucracy, arbitrariness on the part of officials, and a restriction of the flexibility which has hitherto benefited more than one industry. The key to the success of the whole undertaking will therefore be the efficient staffing and preparation of the Inspectorate itself, together with a balanced social dialogue capable of reconciling the interests of business with respect for employee rights.

Implementing the changes described here requires not only legislative courage but also continuous monitoring of the economic effects the reform will have on small and medium-sized enterprises. It is worth remembering that the growth in the competences of supervisory authorities must go hand in hand with growth in their professionalism, in order to avoid a wave of unnecessary court disputes and the paralysis of business activity. Ultimately, the prospects facing the Polish labour market depend on whether the amended provisions become an effective tool for building fair relationships, or merely another bureaucratic challenge. The long-term effects of these measures will shape employment standards and the sense of social security of millions of citizens for years to come.

Bibliography

Government Legislation Centre, file of the draft act: https://legislacja.gov.pl/projekt/12401602/katalog/13153003

Act of 11 March 2026 amending the Act on the National Labour Inspectorate and certain other acts (Journal of Laws of 2026, item 473): https://isap.sejm.gov.pl/isap.nsf/download.xsp/WDU20260000473/O/D20260473.pdf

Opinion of the Association of Labour Inspectors of the Republic of Poland on the draft act amending the Act on the National Labour Inspectorate and certain other acts – draft of 1 September 2025 (No UD283 in the list of legislative and programme works of the Council of Ministers): https://www.siprp.pl/wp-content/uploads/2025/09/opinia-SIPRP_1709.pdf (accessed 4 August 2026)

Trade Unions Forum (FZZ), Comparative analysis of the draft reforms of the National Labour Inspectorate: https://fzz.org.pl/wp-content/uploads/2026/02/Analiza-porownawcza-projektow-reformy-Panstwowej-Inspekcji-Pracy-.pdf (accessed 4 August 2026)

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

Directive (EU) 2024/2831 of the European Parliament and of the Council of 23 October 2024 on improving working conditions in platform work (OJ EU L of 2024, item 2831).

Directive (EU) 2019/1152 of the European Parliament and of the Council of 20 June 2019 on transparent and predictable working conditions in the European Union (OJ EU L 186 of 2019, p. 105).

Directive 96/71/EC of the European Parliament and of the Council of 16 December 1996 concerning the posting of workers in the framework of the provision of services (OJ EU L 18 of 1997, p. 1, as amended).

Directive (EU) 2018/957 of the European Parliament and of the Council of 28 June 2018 amending Directive 96/71/EC concerning the posting of workers in the framework of the provision of services (OJ EU L 173 of 9 July 2018, p. 16).

Directive (EU) 2019/1937 of the European Parliament and of the Council of 23 October 2019 on the protection of persons who report breaches of Union law (OJ EU L 305 of 26 November 2019, p. 17).

Decision of the Supreme Court of 25 November 2025, I USK 433/24, LEX No 3945328.

European Labour Authority: https://www.ela.europa.eu/en

K. Lisiecka, M. Maksis, K. Polewka-Włoch, A. Szulc, Nowe uprawnienia PIP po zmianach 2026. Umowy cywilnoprawne w praktyce [The New Powers of the PIP after the 2026 Changes: Civil-Law Contracts in Practice], Warsaw 2026.

https://rpms.pl/umowy-b2b-zlecenie-2026-nowe-przepisy-ryzyko-uop

[1]A nationwide organisation associating primarily employees of, and labour inspectors connected with, the National Labour Inspectorate.

[2]One of the largest and most important trade union confederations in Poland, representing the interests of workers from various industries and sectors of the economy in dialogue with governments and employers.

[3]A nationwide, cross-sectoral employers’ association bringing together Polish entrepreneurs (both small and medium-sized firms and large enterprises).

Artykuł The Reform of the Powers of the Polish National Labour Inspectorate (PIP) pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

In legal and commercial practice, the notion of the “no-confidence protocol” (protokół nieufności) displays a specific semantic duality, which draws a clear boundary between the colloquial understanding of corporate institutions and the automated architecture of the ICT systems of the administration of justice. Verification of this term requires, above all, the rejection of any assumption as to its statutory origin, for a thorough dogmatic analysis leads to the unambiguous conclusion that managerial practice and the registration procedure are not one and the same.

Proceeding to a formal legal deconstruction, it must be stated categorically that no institution bearing the name “no-confidence protocol” has been provided for in the provisions of the Commercial Companies Code[1] or in the Act on the National Court Register[2]. Polish company law does not employ this terminology at any stage of the regulation of the functioning of the governing bodies of companies, which precludes treating this document as an autonomous instrument of a substantive or procedural character. This phenomenon illustrates a broader tendency for divergences to arise between the language of the law and the language of legal practice, brought about by the implementation of advanced IT solutions in the registry judiciary.

This conceptual duality means that the no-confidence protocol functions in practice as a category of a technical and systemic character, whose erroneous identification with institutions of corporate law results from the penetration of IT jargon into the language of business practice. The actual source of this concept, which appears in the Repository of Registry Files (Repozytorium Akt Rejestrowych), lies exclusively in the sphere of the technical operation of the ICT systems of the Ministry of Justice[3]. When applications for the entry or amendment of data concerning persons appointed to company bodies are examined, the IT System of the National Court Register automatically generates an internal verification document.

This protocol records the course of an algorithm which performs automated queries to external databases, including the National Criminal Register and the Register of Insolvent Debtors[4]. The purpose of this operation is the digital confirmation of the absence of negative circumstances precluding the performance of the function of a member of the management board or the supervisory board. Examples of such bars are the prohibitions set out in Article 18 § 2 of the Commercial Companies Code, whereby the legislature introduces an absolute prohibition on the holding of office by persons convicted by a final judgment of specified economic offences, fiscal-penal offences and offences against the credibility of documents.

A different matter is the practice of business, in which the term functions as a colloquial description of measures connected with the formal expression of a lack of confidence by the company’s owners. Shareholders sometimes use this label for resolutions of shareholders’ meetings concerning the refusal to grant discharge (absolutorium) or the adoption of a vote of no confidence in members of the management board[5]. Although these acts produce significant effects within the company and frequently lead to the immediate removal of managers, in the light of the settled case law of the Supreme Court they constitute an expression of the corporate rights of the shareholders, and not the implementation of an external registration procedure[6].

Summing up the issue under analysis, the conclusion may be formulated that the “no-confidence protocol” is a purely systemic concept which has become widespread in business language, causing misunderstandings among auditors and legal analysts. The contemporary verification of digital registry files requires a clear separation of the technical sphere of automated queries from the legal acts performed by the company’s bodies. Ignoring this distinction could lead to erroneous conclusions concerning the legal status of management board members, whereas the presence of this document in the Court Registers Portal is merely technical proof that the registry court has duly carried out the automated check required by law.

Origins and Circumstances of the Creation of Documents of a No-Confidence Character

An analysis of the reasons and circumstances in which traces of documents described as a “no-confidence protocol” appear in a company’s corporate history requires a clear separation of two distinct planes: the technological-registration plane and internal corporate practice. The first of these, systemic in character, is directly connected with the mechanisms of the operation of the Court Registers Portal and of the ICT system serving the National Court Register (SI KRS). When applications are filed – on forms KRS-Z3 or KRS-W3 – for the entry of newly appointed members of company bodies or for a change in the existing representation, the system obliges users to provide the candidates’ personal data, including their PESEL number. At this stage an automatic, systemic exchange of information takes place between the databases of the Minister of Justice and the National Criminal Register. This procedure serves the obligatory verification of the absence of convictions for the offences specified in Article 18 § 2 of the Commercial Companies Code, which constitute an absolute negative condition for holding office in the bodies of companies. The presence of a phrase such as “no-confidence protocol” in system metadata or in the history of operations on the files of a given registration case therefore constitutes a technical trace of the automated verification of credibility, of a clean criminal record or of internal ICT security algorithms, and not a physical document produced by participants in economic transactions.

Documents of this character have an entirely different origin where their source is sought on the corporate plane, where they take a material form, filed in the registry records by the company itself as annexes to applications. The circumstances of their preparation are almost invariably inseparably linked with deep decision-making, reputational and structural crises within the ownership structure, reflected in specific provisions of the Commercial Companies Code. The most frequent background is decision-making paralysis of the deadlock type, occurring in companies with a symmetrical balance of ownership power (e.g. a 50/50 shareholding structure), where the impossibility of forming a majority prevents the adoption of key resolutions, and one faction seeks formally to document the loss of confidence in, and the failings of, managers associated with its opponents. An equally dynamic catalyst is the detection of managerial abuse (malpractice), such as acting to the company’s detriment, the siphoning-off of assets through related-party transactions on non-market terms (tunneling) or gross breaches of budgetary discipline. In such circumstances, minority shareholders – exercising the control rights arising under Article 212 of the Commercial Companies Code – or supervisory bodies acting under Article 219 of the Code[7] initiate verification procedures, which often culminate in written reports or protocols expressly articulating charges and a lack of confidence. This situation is compounded by disputes over defects in the representation itself, in which the effectiveness of the appointment or removal of management board members is challenged, transferring the conflict from the company’s boardroom to the courts.

In the light of the settled line of Supreme Court case law, acts expressing shareholders’ lack of confidence in managers – most frequently manifested in a refusal to grant discharge – constitute a sovereign expression of the corporate rights of the company’s owners, and not the implementation of an external registration procedure. Of key importance in this respect is the judgment of the Supreme Court[8] in which the Court subjected the legal character of a failure to grant discharge to detailed analysis. Although the Supreme Court came out unequivocally against the concept of so-called negative resolutions, indicating that the non-adoption of a resolution granting discharge, for want of the required majority of votes “in favour”, is not equivalent to the formal adoption of a resolution “refusing” discharge, it simultaneously recognised the significance of that act in the sphere of facts and of public perception. A refusal to grant discharge, irrespective of the technical-legal form in which it occurs, is in commercial practice treated as a corporate vote of no confidence. The Supreme Court emphasised that this decision falls within the sphere of the discretionary power of the shareholders and rests on their subjective economic and legal assessment of the manager’s activity, provided that it does not bear the hallmarks of manifest harassment or bad faith, which could then justify protection of the aggrieved party’s personal rights on the basis of Article 24 of the Civil Code[9].

An independent substantive assessment of the phenomenon under analysis leads to the conclusion that the concept of the “no-confidence protocol” plays a dual, though asymmetrical, role in commercial practice. On the one hand, it constitutes striking evidence of the progressive digitisation and automation of registration procedures, in which algorithmic mechanisms generate systemic traces in order to protect the certainty of legal transactions and to verify the requirements of Article 18 § 2 of the Commercial Companies Code. On the other hand, in practical terms, the materialisation of such documents in a company’s files is a symptom of deep corporate pathology and of the collapse of the fiduciary relationship between capital and management. These documents – whether they take the form of minutes of a stormy supervisory board meeting or of an audit report of minority shareholders – are instruments of corporate warfare of high business significance. They mark the moment of transition from the amicable management of the company to open legal conflict, often preceding the immediate removal of management board members for important reasons (Article 203 § 1 or Article 368 § 4 of the Commercial Companies Code) and opening the way to actions for damages against managers under Article 293 or Article 483 of the Code[10]. For external participants in commercial dealings – including financial institutions and contractual counterparties – the disclosure of such documents, or of their systemic equivalents in the register, constitutes an unambiguous warning signal of elevated operational risk and of the potential loss of decision-making capacity by the entity concerned.

Significance in the Due Diligence Process – Warning Signals

The identification of a reference to, or of a document described as, a “no-confidence protocol” in the course of a professional due diligence review of a counterparty requires the auditor immediately to implement differentiated risk-verification paths, determined by the dual character of this phenomenon. The key task at the stage of documentation analysis is the precise separation of a routine systemic-technical trace from a material corporate document, since each of them generates a diametrically different impact on the stability and security of the planned contractual relationship.

Verification on the systemic plane, revealing the presence of a “no-confidence protocol” record in the metadata of the IT System of the National Court Register (SI KRS) or in the Repository of Registry Files, is, as a rule, a neutral signal. For the entity examining the counterparty’s legal condition, such an entry merely confirms the correct, automatic activation of control algorithms by the IT architecture of the Ministry of Justice, consisting in the querying of external databases, including the National Criminal Register and the Register of Insolvent Debtors. This activity serves the performance of the statutory obligation to examine the conditions of Article 18 § 2 of the Commercial Companies Code, guaranteeing that the persons notified for entry as the company’s representatives are not subject to a prohibition on serving as officers of company bodies. Provided that the registration proceedings ended with a final entry of the data applied for, a technological trace of this kind carries no independent transactional risk and attests merely to the completion of the standard verification procedure.

A completely different, critical significance attaches to the identification, in the registry files or in internal documentation, of a genuine corporate document reflecting a formalised conflict within the company’s ownership structures. Such a document may take the form of minutes of a meeting of the supervisory board or of the shareholders’ meeting at which charges against the management were openly formulated or discharge was refused. In this respect a direct link emerges with the interpretation adopted by the Supreme Court[11]: while the Court indicated that the non-adoption of a discharge resolution for want of the requisite majority of votes does not create a legal fiction in the form of a “negative resolution” refusing discharge, it simultaneously confirmed that in economic reality and in public perception such an act is unequivocally identified with a vote of no confidence. For the entity conducting due diligence, the discovery of a material protocol of such content constitutes a stark warning signal of the breakdown of the relationship of trust between capital and management, which implies three fundamental categories of legal and business risk.

The first of these is the risk of instability of the counterparty’s bodies and of uncertainty as to its representation. A deep corporate crisis, of which a no-confidence protocol is an expression, most frequently precedes or directly triggers sudden changes in the composition of the management board on the basis of Article 203 § 1 or Article 368 § 4 of the Commercial Companies Code. Contracts and declarations of will made by the company in such an unstable transitional period are highly exposed to subsequent contestation by warring shareholder factions. Litigation opponents may seek to undermine the effectiveness of legal acts by invoking defects in the convening of the bodies, the absence of due authority of the officers to act, or breaches of internal rules of representation, which strikes directly at the certainty of commercial transactions and at the legal security of the other party to the contract.

The second consequence is the direct risk of the company becoming entangled in lengthy and costly court proceedings. The existence of a corporate no-confidence document almost invariably heralds the imminent commencement of actions for the revocation or the declaration of the invalidity of resolutions of the owners’ bodies (Articles 249 and 252 of the Commercial Companies Code), the bringing of actions for damages against managers under Article 293 or Article 483 of the Code on account of alleged acts to the company’s detriment (malpractice or tunneling), and, in extreme cases, an application for the appointment of a court-appointed guardian (kurator) on the basis of Article 42 of the Civil Code. The prospect of years of litigation destabilises the counterparty’s organisational structure and generates risk striking at its market reputation.

The consequence of the above phenomena is the third plane, namely financial and operational risk. Ownership conflicts of an intensity requiring formalisation in the form of protocols of lack of confidence absorb key capital and the operational attention of the management, diverting it from the day-to-day running of the enterprise. This results in a drastic decline in the timeliness of contract performance, decision-making paralysis with respect to key investments, and loss of financial liquidity. In addition, where allegations of asset-stripping are confirmed, there is a real threat that security for damages claims will be established over, or enforcement levied against, key components of the counterparty’s assets, which may prevent it from performing its reciprocal obligations.

An independent substantive assessment of the phenomenon leads to the conclusion that the presence of the “no-confidence protocol” motif in the course of due diligence constitutes one of the most demanding qualitative diagnostic indicators. While its systemic dimension is merely a technological testimony to the transparency of the register and to the performance of its security procedures, its corporate dimension exposes a structural pathology threatening the legal existence of the enterprise. For a professional participant in commercial dealings to ignore a genuine, material vote-of-no-confidence document constitutes gross negligence, since in market reality signals with a stronger warning charge are rarely encountered. A correct interpretation of this phenomenon permits a strategic decision to renegotiate the terms of the transaction, to implement additional legal safeguards (e.g. bank guarantees or a submission to enforcement), or – where the conflict escalates – to withdraw from the contract altogether in order to protect one’s own capital.

The Judicial Perspective: Evidentiary and Practical Significance

An analysis of the adjudicative practice of the common courts and of the Supreme Court points to a highly pragmatic approach of the judiciary to documents described in commercial practice as “no-confidence protocols”. The term itself rarely appears directly in the operative parts of judicial decisions, since it is not an institution of statutory law. The courts treat it as an element of the description of the facts, reconstructed on the basis of the nomenclature adopted by the parties to the corporate dispute themselves. Of key importance to adjudicating panels, however, are the substantive content of such a statement and the procedural rigour attending its preparation, which are subject to meticulous verification in the course of specific contentious and registration proceedings.

The most frequent procedural background against which documents reflecting a vote of no confidence are assessed by the courts are cases for damages on account of the premature or defective removal of a management board member from office. Such a situation arises where a management contract or the company’s articles – taking advantage of the non-mandatory character of the statutory provisions – make the manager’s removal from the body before the end of the term of office conditional upon the existence of important reasons. In disputes of this type, the courts uniformly hold that a loss of confidence may not be of a purely subjective, arbitrary or casuistic character. In order to constitute an effective ground for terminating the legal relationship, or to justify a refusal of discharge, it must rest on objectively verifiable factual premises. In this context, a protocol containing a formalised expression of a lack of confidence becomes key evidentiary material, which the court assesses through the prism of the principle of the free evaluation of evidence. The judges examine whether the charges formulated in the protocol – such as mismanagement, failure to implement the strategy or gross breaches of budgetary discipline – were grounded in truth and whether they objectively gave the shareholders justified grounds for severing the corporate bond with the manager. This corresponds to the interpretative line presented, inter alia, in the judgment of the Supreme Court[12], in which it was indicated that although a refusal to grant discharge or a removal from office falls within the sphere of the sovereign and discretionary ownership rights of the shareholders, the motivation behind those acts may not bear the hallmarks of manifest harassment or bad faith, failing which the way is open to legal protection for the manager.

Alongside the strictly contentious sphere, the documents under discussion play an important role before the registry courts, where the procedural correctness of their preparation is examined. Registry courts, acting within the limits of their statutory powers of cognisance, examine whether the documents attached to an application demonstrate conformity, as to form and content, with the provisions of law. If a protocol reflecting a deep conflict within the company displays formal defects – such as the absence of required signatures, the improper convening of a meeting, failure to satisfy quorum requirements or the exclusion of a shareholder from the vote – the registry court refuses to enter the changes in the structure of the bodies. This temporarily sanctions and freezes the existing balance of power in the company’s representation, paralysing the personnel effects of the decisions taken.

An independent substantive assessment of this issue leads to the conclusion that a protocol of a no-confidence character performs the function of a kind of bridge between the dynamic, often emotional, sphere of internal ownership conflict and the formalised world of judicial procedure. Its evidentiary significance is fundamental, since it is precisely this document that verifies the shareholders’ intentions at the critical moment of corporate crisis. From the point of view of commercial practice, a defectively prepared protocol, or one based on unverified charges, becomes a powerful source of litigation risk for the company, opening the way for the removed manager to bring effective damages claims. For professional auditors and business advisers in the course of due diligence, an analysis of the manner of drafting and of the formal correctness of such documents constitutes a key element of the assessment of whether the changes in the company’s bodies will stand up in court, or whether they will be challenged, generating a prolonged state of legal uncertainty for the entity.

Practical Conclusions and Verification Recommendations

The identification, in the course of a due diligence procedure, of a trace of a document functioning as a “no-confidence protocol” obliges the analyst immediately to implement an in-depth audit verification algorithm aimed at precisely separating systemic risks from corporate ones. The first stage of this process is a formal audit of the registration case file, allowing it to be established whether the reference to this concept is merely a standard, automatically generated record of verification in the National Criminal Register pursuant to Article 18 § 2 of the Commercial Companies Code. If the case has been finally concluded with the entry of the newly appointed members of the representative body, and the record itself is of a purely technical, ICT character within the structure of SI KRS, that circumstance generates no independent transactional risk for the planned agreement. It should also be borne in mind that, in the process of verifying registry files, analysts may encounter judicial decisions of a purely procedural character; such acts must be clearly distinguished from rulings of a substantive-law character, which directly touch upon the essence of the functioning of the company’s bodies and affect the security of commercial transactions.

An entirely different audit procedure is activated where the examination shows that the document reflects a genuine ownership dispute and was physically filed by the company itself or by its shareholders. In such a scenario, the key step is to demand the production of the full corporate documentation directly from the counterparty, including in particular a copy of the notarial minutes of the relevant shareholders’ meeting or general meeting, together with the attendance list, the structure of the distribution of votes and the exact wording of any dissenting opinions filed. This corresponds with the practice arising from the judgment of the Supreme Court (case no. II CSKP 1330/22), which pointed to the economic and structural significance of a failure to grant discharge as an act directed at managers. A full analysis of the content of such a document allows the auditor to reconstruct the real causes of the corporate rift and to assess the substantive merits of the charges of mismanagement or abuse that have been raised.

The next, obligatory step within the verification algorithm is a comprehensive examination of pending disputes. A professional analyst should obtain from the counterparty’s incumbent management board a written statement, made under full liability in damages, confirming the absence of pending court proceedings aimed at challenging resolutions of the owners’ bodies. This verification must expressly cover claims for the revocation or the declaration of the invalidity of resolutions on the basis of Articles 249 and 252 of the Commercial Companies Code in relation to limited liability companies, and Articles 422 and 425 of the Code in the case of joint-stock companies. In parallel, it is advisable to conduct an independent search in the commercial divisions of the competent regional courts for any actions for the payment of damages, or for the shaping of a legal relationship, brought by former management board members removed in an atmosphere of conflict.

Where the steps taken reveal the existence of a genuine, smouldering corporate conflict, and business considerations imperatively require cooperation with the counterparty, it is necessary to implement radical risk-mitigation mechanisms through the introduction of specific contractual clauses. In such circumstances, the client’s legal interest is protected by requiring joint representation at the conclusion of the contract, by obtaining written statements from all warring shareholder factions confirming the full authority of the current management board, or by establishing joint and several security of a personal character. For example, a suretyship given by the natural persons controlling the company, or a submission to enforcement under Article 777 of the Code of Civil Procedure, makes the effective pursuit of claims independent of decision-making paralysis or of any degradation of the asset structure of the counterparty itself.

A substantive assessment of the issue under discussion leads to the conclusion that the implementation of a structured algorithm for the verification of the “no-confidence protocol” elevates the due diligence procedure from the level of a passive examination of documentation to the rank of a strategic predictive tool. The presence of a material vote of no confidence in a company’s history is rarely an incident of a purely historical character; most often it is a symptom of deep structural dysfunction which will, in all likelihood, materialise in the operational sphere. Consequently, the diligent performance of the audit recommendations described above not only protects the examining entity against the risk of concluding a contract affected by legal defects or by the sanction of invalidity, but also supplies invaluable asymmetric knowledge, enabling the renegotiation of the pricing terms of the transaction or the securing of above-standard safeguards for the performance of the obligation.

[1]Act of 15 September 2000 – Commercial Companies Code (consolidated text: Journal of Laws of 2024, item 18, as amended).

[2]Act of 20 August 1997 on the National Court Register (consolidated text: Journal of Laws of 2025, item 869, as amended).

[3]Technical specification of the ICT systems of the Department of Informatisation and Court Registers of the Ministry of Justice.

[4]Article 21a of the Act of 20 August 1997 on the National Court Register (consolidated text: Journal of Laws of 2025, item 869, as amended) in conjunction with the provisions of the Act of 24 May 2000 on the National Criminal Register (consolidated text: Journal of Laws of 2024, item 276, as amended).

[5]Judgment of the Supreme Court of 10 November 2022, case no. II CSKP 1330/22.

[6]Decision of the Supreme Court of 10 January 2019, case no. II CSK 345/18.

[7]Act of 15 September 2000 – Commercial Companies Code (consolidated text: Journal of Laws of 2024, item 18, as amended), Articles 212 and 219.

[8]Judgment of the Supreme Court of 10 November 2022, case no. II CSKP 1330/22.

[9]Act of 23 April 1964 – Civil Code (consolidated text: Journal of Laws of 2025, item 1071, as amended).

[10]Act of 15 September 2000 – Commercial Companies Code (consolidated text: Journal of Laws of 2024, item 18, as amended), Article 203 § 1, Article 368 § 4, Article 293 and Article 483.

[11]Judgment of the Supreme Court of 10 November 2022, case no. II CSKP 1330/22.

[12]Judgment of the Supreme Court of 10 November 2022, case no. II CSKP 1330/22.

Artykuł Legal Status and Definition of the Concept of the “No-Confidence Protocol” (protokół nieufności) pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

I. Introduction

The longevity economy is commonly estimated at several hundred billion dollars worldwide and continues to grow rapidly.[1] Venture capital flows into biotechnology companies that promise to slow, halt, or reverse aging. Clinics from Zurich to Seoul to Miami sell “longevity programmes.” Supplement brands attach the word “longevity” to products ranging from NAD precursors to collagen powders. And yet the word “longevity” does not exist as a legal category in European Union law. There is no definition, no register, no certification scheme, and no authority empowered to decide what may or may not describe itself as a longevity product, a longevity clinic, or a longevity company.

The gap between an enormous market and a non-existent legal definition is precisely where legal risk accumulates. Where there is no definition, there is classification risk, claims risk, and enforcement risk. There is also a phenomenon that this article, borrowing from the vocabulary of environmental marketing, calls “longevity washing.”

This article examines the longevity industry through the lens of EU regulatory law. The EU framework governs a market of some 450 million consumers and, because it is horizontal, sector-specific, and increasingly enforcement-oriented, it offers a useful point of comparison for practitioners in the United States, Asia, and Latin America. Part II explains why “longevity” is not a legal category and introduces the borderline-product problem that follows from that fact. Part III develops the concept of longevity washing and its consumer-facing, unfair-commercial-practice, and investor-facing dimensions. Part IV addresses the regulation of artificial intelligence in longevity research and diagnostics, including the revised timeline of the AI Act, the position of the European Medicines Agency on AI in the medicinal-product lifecycle, and the data-protection architecture formed by the GDPR and the European Health Data Space. Part V maps the sectoral “hot spots” across pharmaceuticals, medical devices, cosmetics, and food supplements. Part VI draws conclusions for practitioners.

II. “Longevity” in the Eyes of the Law: The Borderline-Product Problem

In EU regulatory law, aging is not a disease. The European Medicines Agency has never approved a medicinal product with “aging” as a therapeutic indication, and under Directive 2001/83/EC a medicinal product is defined by reference to the treatment or prevention of disease or the restoration, correction, or modification of physiological functions.[2] The position in the United States is comparable. The long-running debate around the Targeting Aging with Metformin (TAME) trial illustrates the point: its designers had to negotiate a trial protocol with the Food and Drug Administration that targets aging indirectly, through a composite of age-related diseases, precisely because “aging” itself is not an approvable indication.[3]

The consequence is profound. A company cannot lawfully market a product “to treat aging.” The entire longevity industry therefore operates in the space of indirect claims: “supports healthy aging,” “promotes cellular health,” “extends healthspan.” Each of those phrases pushes the product toward one of several very different regulatory regimes.

This is the borderline-product problem, and it is the single most important legal concept in the sector. The same molecule – an NAD precursor, for example – may be: a medicinal product, if it is presented as treating or preventing disease or exerts a significant pharmacological, immunological, or metabolic effect; a food supplement under Directive 2002/46/EC;[4] a novel food requiring pre-market authorisation under Regulation (EU) 2015/2283;[5] a cosmetic under Regulation (EC) No 1223/2009, if applied to the skin with a claim limited to appearance;[6] or a medical device. The last category is broader than it may appear: since the Medical Device Regulation (MDR) came into application, products listed in its Annex XVI – dermal fillers, certain energy-based aesthetic equipment – are regulated as devices even where they have no medical purpose at all.[7]

Classification does not follow the label a company chooses. It follows presentation, composition, and route of administration. The Court of Justice of the European Union has held for four decades that a product is a medicinal product “by presentation” where it is described or recommended as having properties for treating or preventing disease, irrespective of whether it actually has those properties; and that classification “by function” turns on a case-by-case scientific assessment of the product’s actual pharmacological, immunological, or metabolic action, not on the manufacturer’s intention.[8] Marketing language alone can therefore reclassify a product and, with it, the entire compliance burden: marketing authorisation, good manufacturing practice, pharmacovigilance, and advertising restrictions.

When a client says “we are a longevity company,” the first question counsel should ask is: which regime are you actually in? “Longevity” is a marketing word. The law knows only medicines, devices, foods, and cosmetics.

III. Longevity Washing: When the Claim Outruns the Science

Greenwashing – claiming environmental virtue one does not have – is a familiar concept. Longevity washing is its younger sibling: labelling a product, a clinic, a research programme, or an entire company as “longevity-focused” when the science, the indication, or the business model does not support that description. The practice matters legally for three reasons.

A. Consumer-Facing Claims

Under the Health Claims Regulation, any health claim made on a food – and food supplements are foods – must be specifically authorised and listed in the EU register.[9] Claims such as “slows aging,” “supports cellular repair,” or “extends lifespan” are not authorised health claims. National authorities in the EU and the United Kingdom have treated phrases such as “cellular repair” and even “healthy aging” as impermissible where they imply a physiological mechanism or a reduction of disease risk. A substantial part of the longevity supplement market is thus operating with claims that would not survive enforcement scrutiny.

B. Unfair Commercial Practices and the Greenwashing Template

The second layer is the general prohibition of misleading commercial practices under the Unfair Commercial Practices Directive (UCPD).[10] Here the EU has recently built an entire enforcement architecture against greenwashing that repays close study by anyone advising on health claims. The Empowering Consumers for the Green Transition Directive amends the UCPD to prohibit generic environmental claims – “eco-friendly,” “climate-neutral” – unless the trader can demonstrate recognised excellent environmental performance, and it adds a series of specific greenwashing practices to the UCPD’s blacklist. Its provisions apply across the Union from 27 September 2026.[11]

The Directive should be read as a template. If “climate-neutral” now requires substantiation, it is a very short doctrinal step for regulators and courts to demand the same of “clinically proven to slow biological aging.” The methodology of anti-greenwashing enforcement – generic claim, absence of substantiation, misleading omission – maps directly onto longevity claims. The UCPD’s general clauses on misleading actions and omissions already provide the legal basis; what the green-claims reform supplies is a worked example of how those clauses can be operationalised against vague virtue-signalling.

C. Investor-Facing Washing

The third dimension is of particular relevance to transactional lawyers. Start-ups increasingly brand themselves as “longevity biotech” because that is where the capital is, even when the underlying asset is a conventional dermatology product or a wellness application. For investors, this creates due-diligence risk: is the “longevity platform” in fact a regulated medicinal pipeline, with all the timeline and cost that implies, or a supplement business one enforcement letter away from relabelling? Misclassification discovered after investment is a valuation event. For the company, aggressive longevity framing in fundraising materials can generate liability under securities and misrepresentation rules in multiple jurisdictions. In due diligence on a longevity start-up, the regulatory classification memorandum is not an annex to the deal; it is the deal.

IV. Artificial Intelligence in Longevity Research and Diagnostics

The longevity field is arguably the most AI-dependent sector in the life sciences, for a simple reason: aging is not one target but thousands of interacting biological processes. A geroprotective drug cannot be designed the way an antibiotic is designed. The entire aging phenotype must be modelled, and that means machine learning on large multi-omic datasets. AI is not an add-on in longevity; it is the core methodology. That raises three distinct regulatory questions under EU law.

A. The AI Act

Regulation (EU) 2024/1689, the AI Act, is the most comprehensive horizontal AI regulation in force anywhere and applies to almost every AI system used in the longevity value chain.[12] It adopts a risk-based structure – prohibited practices, high-risk systems, transparency obligations, and minimal-risk systems – and the critical category for this sector is high-risk. An AI system that is itself a medical device, or a safety component of one, under the MDR or the In Vitro Diagnostic Regulation, and that is subject to third-party conformity assessment under those instruments, is high-risk under Article 6(1) and Annex I of the AI Act.[13] This captures, for example, an AI tool that estimates a patient’s “biological age” from a blood panel and uses that estimate to guide a clinical intervention; an AI-driven diagnostic that predicts age-related disease risk; and the software layer of many longevity-clinic platforms once it makes a diagnostic or prognostic claim.

High-risk status is not a prohibition; it is a compliance regime. It requires a risk-management system, data governance with representative and bias-tested training data, technical documentation, logging, human oversight, transparency to deployers, accuracy, robustness and cybersecurity, a conformity assessment, and post-market monitoring.[14] For most longevity start-ups this is a wholly new compliance layer on top of MDR conformity assessment, and the two must be coordinated.

The timeline has recently changed, and practitioners should update their advice accordingly. The original AI Act applied the high-risk obligations from 2 August 2026 for stand-alone systems listed in Annex III and from 2 August 2027 for AI embedded in products covered by Annex I, including medical devices. The Digital Omnibus on AI, adopted as Regulation (EU) 2026/1744 and in force since 27 July 2026, defers those dates to 2 December 2027 and 2 August 2028 respectively.[15] The deferral does not lighten the substantive regime; the prohibitions in Article 5, the obligations of providers of general-purpose AI models, and the transparency obligations of Article 50 (which apply from 2 August 2026) are unaffected. For longevity companies whose AI is a medical device, the practical effect is a compliance window of roughly two years, during which harmonised standards and notified-body capacity are expected to mature. The window is an opportunity to design AI Act compliance into the MDR technical file from the outset rather than to defer the exercise.

B. AI in Drug Discovery and the European Medicines Agency

The position for drug discovery is more nuanced. An AI model used purely internally to identify candidate molecules – the classic “AI for drug discovery” use case that dominates longevity-biotech pitch decks – is generally not, in itself, a high-risk AI system under the AI Act, because it does not make or materially influence decisions about individual patients. The regulatory question shifts instead to the European Medicines Agency. The EMA’s Reflection Paper on the use of AI in the medicinal product lifecycle sets out the Agency’s expectations from discovery through pharmacovigilance.[16] Its core message is that the burden of proof rests on the sponsor to demonstrate that the AI model is fit for its regulatory purpose. That entails documented data provenance, model validation, a degree of explainability appropriate to the stage of development, and lifecycle management of the model itself.

If a longevity company tells its investors that “our AI discovered this molecule,” the regulator will eventually ask to see the model, the training data, and the validation. Vague AI claims in fundraising materials do not survive regulatory scrutiny and, increasingly, do not survive investor due diligence either.

C. Data: The GDPR and the European Health Data Space

The third question concerns data, and it is here that longevity becomes uniquely difficult. Aging research is inherently longitudinal and multi-omic: genomic, epigenetic, and proteomic data, wearable data, and medical records, ideally across decades. Under the General Data Protection Regulation this is special-category data – health, genetic, and biometric data – and the available lawful bases are narrow.[17] The scientific-research provisions of Article 9(2)(j) and Article 89 assist, but they are implemented differently in each Member State, which produces real fragmentation for pan-European longevity cohorts.

The European Health Data Space Regulation adds a second layer.[18] It creates, for the first time, a mandatory EU-wide framework for the secondary use of electronic health data, administered by health-data-access bodies in each Member State. For longevity research the EHDS is potentially transformative, because it opens access to population-scale health data for research purposes. It also imposes strict conditions: purpose limitation, processing only within secure processing environments, a prohibition on re-identification, and specific prohibitions on using the data for advertising or marketing, for decisions detrimental to individuals such as insurance or employment decisions, or for developing products harmful to health.[19] The secondary-use chapter applies from March 2029 for most data categories, but the lead time is short in the context of longitudinal cohorts. Longevity companies that intend to monetise health data – and many do – should be mapping their data flows against the EHDS regime now.

V. Sector-by-Sector Regulatory Hot Spots

Four sectoral silos together make up the longevity industry, and the same start-up frequently straddles two or three of them without realising it. In each silo, specific “hot spots” – advertising, manufacturing standards, registration, and market withdrawal – recur in practice.

A. Pharmaceuticals

True longevity drugs – rapamycin analogues, senolytics, geroprotectors – sit squarely within Directive 2001/83/EC and its implementing rules, and will continue to do so under the new pharmaceutical package once it applies. Three practical hot spots stand out. First, off-label prescribing of rapamycin and metformin for longevity indications is widespread in private clinics across Europe. Off-label prescribing as such is a matter of clinical freedom, but off-label promotion by or on behalf of the marketing-authorisation holder is prohibited: advertising must conform to the summary of product characteristics, and advertising of unauthorised products or indications is not permitted.[20] Longevity clinics are increasingly the vehicle through which promotional messaging reaches patients, and enforcement authorities have begun to take notice. Second, good manufacturing practice: anything presented as a medicine must be manufactured under an EU manufacturing authorisation and EU GMP, with qualified persons and full traceability; compounding pharmacies producing “longevity peptides” have become a significant enforcement target. Third, unlicensed importation: a large part of the longevity-peptide market operates through personal importation and grey channels, and customs enforcement across the EU has visibly tightened.

B. Medical Devices

The MDR has fundamentally reshaped this space. Aesthetic devices – dermal fillers, energy-based devices used in longevity clinics, at-home devices marketed for “cellular rejuvenation” – are caught by Annex XVI even without a medical purpose. Three hot spots follow. Classification: many longevity wearables and applications become software as a medical device once they make a diagnostic or prognostic claim. Advertising: Article 7 of the MDR prohibits misleading claims and specifically prohibits ascribing to a device functions and properties it does not have, which operates in practice as a codified anti-washing rule for devices.[21] Registration and vigilance: the first four modules of the EUDAMED database – actor registration, UDI/device registration, notified bodies and certificates, and market surveillance – have been mandatory since 28 May 2026, with legacy devices to be registered by 28 November 2026; the vigilance module is not yet mandatory, so serious incidents and field safety corrective actions continue to be reported to national competent authorities.[22] Market withdrawals happen: a device sold as a “cellular biomarker analyser” without valid clinical evidence can be removed from the market by any national competent authority, and the MDR’s market-surveillance coordination means that one withdrawal can cascade across the Union.

C. Cosmetics and Aesthetic Products

The Cosmetics Regulation governs anti-aging creams, serums, and topical actives. Two hot spots dominate. The first is the Product Information File: the responsible person must hold a complete file including a safety assessment and substantiation of every claim. The second is the Claims Regulation, Commission Regulation (EU) No 655/2013, with its six common criteria: legal compliance, truthfulness, evidential support, honesty, fairness, and informed decision-making.[23] A claim such as “reverses skin aging at the cellular level” fails the evidential-support criterion unless a robust dossier stands behind it. Enforcement is largely national, but the Cosmetic Products Notification Portal is EU-wide, and non-compliant products are routinely notified through the Safety Gate rapid-alert system.

D. Food Supplements and Novel Foods

It is in the food and supplement space that the longevity industry is most creative and most exposed. Directive 2002/46/EC governs vitamins and minerals in supplements; everything else falls under the general food law of Regulation (EC) No 178/2002[24] and, critically, the Novel Food Regulation. Nicotinamide mononucleotide (NMN), one of the most heavily marketed longevity molecules, is the paradigmatic case. NMN is treated as a novel food in the EU, which means it cannot lawfully be placed on the market as a supplement without pre-market authorisation by the Commission following an EFSA safety assessment. Nicotinamide riboside received such an authorisation in 2020.[25] NMN has not – although the picture is now moving. In March 2026 EFSA’s NDA Panel adopted a positive opinion on the safety of β-NMN in food supplements at up to 300 mg per day for adults, and the opinion was published in May 2026.[26] An EFSA opinion is a scientific assessment, not an authorisation; the Commission must still adopt an implementing act adding NMN to the Union list, and where the applicant has requested data protection the authorisation will initially be proprietary to that applicant’s specification. Until that act is adopted, NMN products on the shelves of European longevity clinics remain unauthorised novel foods. That is not a grey area; it is a non-compliant market operating in plain sight, and it is the kind of enforcement pipeline that eventually produces high-profile withdrawal actions. Once the Health Claims Regulation is added to the picture, the conclusion is unavoidable: the food and supplement leg of the longevity industry is the most legally exposed part of the entire sector.

VI. Conclusions

The longevity industry is a genuine scientific and commercial phenomenon, and much of the underlying research is serious. But lawyers advising companies, investors, clinicians, and, increasingly, regulators in this space must see it clearly: “longevity” is not a legal category. It is a marketing frame layered on top of four very different regulatory regimes. Three practical conclusions follow for anyone whose product touches European consumers.

First, always begin with classification. The word “longevity” conveys nothing about the applicable regime. Ask what the product is, how it is presented, and what it does, and let the regulatory regime follow from those answers.

Second, treat longevity claims the way green claims are now treated. The direction of enforcement is unmistakable. The Empowering Consumers for the Green Transition Directive, applicable from 27 September 2026, gives regulators a sharpened toolkit that will, sooner or later, be pointed at health and wellness claims. Substantiation is no longer optional.

Third, integrate AI Act compliance with sectoral compliance from the outset. For any longevity venture using AI in diagnostics or patient-facing decision support, the AI Act, the MDR, and the GDPR, soon joined by the European Health Data Space, form a single interlocking compliance architecture. The deferral of the AI Act’s high-risk obligations to 2027 and 2028 is a design window, not a reprieve. Retrofitting is expensive; designing for compliance is not.

The longevity industry will keep growing, and regulators will keep catching up. The lawyers who understand both the science and the sectoral map will be the ones clients call first.

[1] Longevity economy estimates vary widely by definition; see, e.g., the periodic longevity-market reports of Bank of America Global Research and McKinsey Health Institute. The figure is offered here as an order of magnitude, not a precise valuation.

[2] Directive 2001/83/EC of the European Parliament and of the Council of 6 November 2001 on the Community Code Relating to Medicinal Products for Human Use, art. 1(2), 2001 O.J. (L 311) 67 [hereinafter Directive 2001/83]. At the time of writing, the new EU pharmaceutical package (a new directive on the Union code for medicinal products for human use and a new regulation on Union authorisation procedures) has been politically agreed and is in the final stage of formal adoption; the substantive rules discussed here are carried over into the new framework, which will apply after a transitional period of approximately two years from entry into force.

[3] See Nir Barzilai et al., Metformin as a Tool to Target Aging, 23 Cell Metabolism 1060 (2016) (describing the design of the Targeting Aging with Metformin (TAME) trial around a composite of age-related diseases).

[4] Directive 2002/46/EC of the European Parliament and of the Council of 10 June 2002 on the Approximation of the Laws of the Member States Relating to Food Supplements, 2002 O.J. (L 183) 51.

[5] Regulation (EU) 2015/2283 of the European Parliament and of the Council of 25 November 2015 on Novel Foods, 2015 O.J. (L 327) 1 [hereinafter Novel Food Regulation].

[6] Regulation (EC) No 1223/2009 of the European Parliament and of the Council of 30 November 2009 on Cosmetic Products, 2009 O.J. (L 342) 59 [hereinafter Cosmetics Regulation].

[7] Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on Medical Devices, Annex XVI, 2017 O.J. (L 117) 1 [hereinafter MDR]; Commission Implementing Regulation (EU) 2022/2346 of 1 December 2022 Laying Down Common Specifications for the Groups of Products Without an Intended Medical Purpose Listed in Annex XVI, 2022 O.J. (L 311) 60.

[8] Case 227/82, Van Bennekom, ECLI:EU:C:1983:354 (establishing that a product is a medicinal product “by presentation” where it is described or recommended as having therapeutic or prophylactic properties, irrespective of its actual efficacy); Case C-319/05, Comm’n v. Germany, ECLI:EU:C:2007:678 (garlic capsules); Case C-140/07, Hecht-Pharma GmbH v. Staatliches Gewerbeaufsichtsamt Luneburg, ECLI:EU:C:2009:5 (classification “by function” requires a case-by-case assessment of pharmacological, immunological or metabolic properties). See also Directive 2001/83, supra note 2, art. 2(2) (in cases of doubt, the medicinal-products regime prevails).

[9] Regulation (EC) No 1924/2006 of the European Parliament and of the Council of 20 December 2006 on Nutrition and Health Claims Made on Foods, arts. 10, 13, 14, 2006 O.J. (L 404) 9 [hereinafter Health Claims Regulation]. The EU Register of nutrition and health claims is maintained by the European Commission.

[10] Directive 2005/29/EC of the European Parliament and of the Council of 11 May 2005 Concerning Unfair Business-to-Consumer Commercial Practices in the Internal Market, 2005 O.J. (L 149) 22 [hereinafter UCPD].

[11] Directive (EU) 2024/825 of the European Parliament and of the Council of 28 February 2024 Amending Directives 2005/29/EC and 2011/83/EU as Regards Empowering Consumers for the Green Transition, OJ L, 2024/825, 6.3.2024 [hereinafter ECGT Directive]. Member States were required to transpose the Directive by 27 March 2026; its provisions apply from 27 September 2026. The separate Green Claims Directive proposal, which would have introduced ex ante verification of explicit environmental claims, was withdrawn by the Commission in 2025 and is not discussed here.

[12] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence, OJ L, 2024/1689, 12.7.2024 [hereinafter AI Act].

[13] AI Act, supra note 12, art. 6(1) & Annex I, § A (listing the MDR and Regulation (EU) 2017/746 on in vitro diagnostic medical devices). The Annex I route applies where the AI system is a product, or a safety component of a product, that is itself subject to third-party conformity assessment under the listed Union harmonisation legislation.

[14] AI Act, supra note 12, arts. 9–15, 17, 43, 72.

[15] Regulation (EU) 2026/1744 of the European Parliament and of the Council Amending Regulation (EU) 2024/1689 as Regards the Simplification of Certain Obligations (“Digital Omnibus on AI”), OJ L, 2026/1744, 24.7.2026 (in force 27 July 2026). The Regulation defers the high-risk obligations for stand-alone (Annex III) systems to 2 December 2027 and for AI embedded in products covered by Annex I, including medical devices, to 2 August 2028; the Article 50 transparency obligations continue to apply from 2 August 2026.

[16] Eur. Medicines Agency, Reflection Paper on the Use of Artificial Intelligence (AI) in the Medicinal Product Lifecycle, EMA/CHMP/CVMP/83833/2023 (adopted Sept. 2024).

[17] 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, arts. 9(2)(j), 89, 2016 O.J. (L 119) 1 [hereinafter GDPR].

[18] Regulation (EU) 2025/327 of the European Parliament and of the Council of 11 February 2025 on the European Health Data Space, OJ L, 2025/327, 5.3.2025 [hereinafter EHDS Regulation]. The Regulation entered into force on 26 March 2025 and applies in stages: most provisions from 26 March 2027, and the core secondary-use chapter from 26 March 2029 (2031 for certain data categories). See id. art. 105.

[19] EHDS Regulation, supra note 18, arts. 50–54 (secure processing environments; permitted and prohibited purposes of secondary use).

[20] Directive 2001/83, supra note 2, art. 87(2) (all parts of advertising must comply with the particulars listed in the summary of product characteristics), arts. 40–53 (manufacturing authorisation and good manufacturing practice) and art. 87(1) (prohibition of advertising of medicinal products without a marketing authorisation).

[21] MDR, supra note 7, art. 7.

[22] Commission Decision (EU) 2025/2371 of 26 November 2025 (declaring the first four EUDAMED modules functional), OJ L, 2025/2371, 27.11.2025; Regulation (EU) 2024/1860 of the European Parliament and of the Council of 13 June 2024, OJ L, 2024/1860, 9.7.2024 (gradual roll-out of EUDAMED). The four modules (actor registration, UDI/device registration, notified bodies and certificates, market surveillance) are mandatory from 28 May 2026; legacy devices remaining on the market must be registered by 28 November 2026. The vigilance and clinical-investigation modules are not yet mandatory, and serious incidents continue to be reported to national competent authorities.

[23] Cosmetics Regulation, supra note 6, arts. 11, 13, 20; Commission Regulation (EU) No 655/2013 of 10 July 2013 Laying Down Common Criteria for the Justification of Claims Used in Relation to Cosmetic Products, Annex, 2013 O.J. (L 190) 31.

[24] Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January 2002 Laying Down the General Principles and Requirements of Food Law, 2002 O.J. (L 31) 1.

[25] Commission Implementing Regulation (EU) 2020/16 of 10 January 2020 Authorising the Placing on the Market of Nicotinamide Riboside Chloride as a Novel Food, 2020 O.J. (L 7) 6.

[26] EFSA Panel on Nutrition, Novel Foods and Food Allergens, Safety of Beta-Nicotinamide Mononucleotide (β-NMN) Pursuant to Regulation (EU) 2015/2283 and the Bioavailability of Nicotinamide from This Source in the Context of Directive 2002/46/EC, 24 EFSA J. e10007 (2026) (opinion adopted 4 March 2026, published 11 May 2026). Authorisation requires a Commission implementing act amending the Union list established by Commission Implementing Regulation (EU) 2017/2470; at the time of writing no such act has been published, and any authorisation granted on the basis of proprietary data will initially benefit only the applicant. See Novel Food Regulation, supra note 5, arts. 12, 26.

Artykuł Longevity Washing, AI and the Regulatory Fault Lines of the Longevity Industry: A European Perspective pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

As part of the American Bar Association (ABA) Section of International Law’s highly successful conference at Mansion House in London, we had the opportunity to take part not only in the substantive programme of the conference, but also in a number of exceptional events accompanying it.

At the heart of the conference were panels addressing some of the most advanced and rapidly developing areas of international law, including artificial intelligence and the law, international finance, M&A transactions, and other key issues shaping the future of the global legal profession.

What makes events organised by the ABA particularly unique, however, is the opportunity to engage with leading practitioners and distinguished legal experts in settings that go far beyond the traditional conference room.

One of the highlights was an exclusive visit to the United Kingdom Parliament, organised by the ABA Section of International Law. Thanks to the kind assistance and hospitality of distinguished British lawyers and experts in the rule of law, we had the privilege of visiting both the House of Lords and the House of Commons and learning more about the history and functioning of the British parliamentary system.

The visit to the House of Commons began in Westminster Hall — one of the most remarkable historic buildings in London, dating back to the 11th century and forming one of the oldest surviving parts of the Palace of Westminster.

For us, this was a truly exceptional experience: an opportunity to combine the substantive exchange of knowledge on the future of international law with direct exposure to one of the world’s most important constitutional and parliamentary institutions.

We are proud to have participated in this outstanding ABA Section of International Law conference and grateful for the opportunity to meet so many distinguished lawyers and legal professionals from the United Kingdom, London, and around the world.

#AmericanBarAssociation #ABA #InternationalLaw #ABALaw #London #LegalConference #InternationalLawyer #LawFirm #Lawyers #LegalProfession #RuleOfLaw #HouseOfLords #HouseOfCommons #Westminster #UKParliament #ArtificialIntelligenceLaw #AILaw #MALaw #MA #CorporateLaw #InternationalBusiness #LegalTech #FutureOfLaw #KieltykaGladkowski #KieltykaGladkowskiLawFirm

Artykuł KGLEGAL Kieltyka Gładkowski – ABA International LC in London | Inside the House of Lords & House of Commons pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

The loss or corruption of data in a business is rarely “just” 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.

Personal Data and Non-Personal Data

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).

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:

• technical, operational, industrial and telemetry data,
• financial and production data,
• source code and design documentation,
• artificial intelligence models together with their training data,
• data on the operation of IT systems.

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.

A Different Centre of Gravity of Protection

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:

• 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;
• 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.

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.

Mixed Data Sets

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.

Liability for Loss or Corruption of Personal Data

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

• a personal data breach (the security incident),
• an infringement of the GDPR by the controller or processor,
• damage suffered by the data subject.

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.

Personal Data Breach

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:

• confidentiality – unauthorised disclosure of, or access to, the data;
• integrity – unauthorised alteration of the data, i.e. any change made by an unauthorised person or an incorrect change made by an authorised person;
• availability – 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).

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

• human error, e.g. a mistake, a lost storage device or a failure to recognise that a breach has occurred;
• inadequate safeguards or procedures;
• 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);
• physical or environmental factors, such as natural disasters and failures of technical infrastructure.

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).

Infringement of the GDPR

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.

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:

• Article 24(1) GDPR – the obligation to implement appropriate technical and organisational measures and to assess risk;
• Article 25(1) GDPR – the obligation to take data protection into account already at the design stage of the processing and throughout the processing itself (privacy by design);
• Article 32(1) GDPR – the obligation to implement measures ensuring a level of security appropriate to the risk, e.g. authentication procedures and appropriate infrastructure and safeguards;
• Articles 33 and 34 GDPR – 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;
• Article 35 GDPR – 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).

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.

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.

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:

• the right of access (Article 15 GDPR) – 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;
• the right to rectification (Article 16 GDPR) – 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;
• the right to erasure (Article 17 GDPR) – 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;
• the right to restriction of processing (Article 18(1)(a) GDPR) – where the data subject contests the accuracy of the data, which is precisely the case when data has been corrupted.

Infringement of the provisions governing data subjects’ 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.

Damage and Liability for Compensation

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)).

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:

• C-300/21 (Österreichische Post) – 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;
• C-687/21 (BL v MediaMarktSaturn Hagen-Iserlohn) – 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;
• judgment of the Warsaw Court of Appeal of 22 June 2023, case no. I ACa 352/23 – the court confirmed that non-material damage within the meaning of Article 82 GDPR also covers distress and the risk of reputational harm.

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.

Liability for Loss or Corruption of Non-Personal Data

Contractual Liability

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.

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:

• Recovery Point Objective (RPO) – 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;
• Recovery Time Objective (RTO) – the maximum acceptable time for restoring operational processes after a disruption.

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:

• prevention (backups, protective systems, maintenance),
• detection (monitoring, security audits),
• remediation (data restoration, activation of the business continuity plan).

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.

Contractual Penalties, Limitation of Liability and Force Majeure

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.

The parties may contractually modify the scope of the provider’s liability, in particular by:

• limiting liability to actual loss, excluding lost profits,
• introducing a monetary cap on damages,
• excluding the statutory warranty for defects.

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.

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.

Other Bases of Liability

Criminal Liability

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’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.

Liability of Members of Company Bodies

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.

Loss or Corruption of Data as a Cybersecurity Incident

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’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:

• ICT risk management – 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.
• Digital resilience and business continuity – 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.
• Oversight of third-party providers – 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.

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.

Conclusion

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:

• personal data – administrative liability arises from an infringement of the GDPR, while liability for compensation depends on the existence of damage and a causal link;
• non-personal data – 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;
• regardless of the type of data – 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.

An appropriate response to the incident, including timely notification of the breach to the competent authorities, may significantly reduce the scope of that liability.

Artykuł Liability for Loss and Corruption of Data pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

Publication date: October 05, 2026

When does a State act like a merchant? A scholarly and practical analysis

I. The restrictive theory of State immunity and the significance of the 2004 Convention

The United Nations Convention on Jurisdictional Immunities of States and Their Property, adopted by General Assembly resolution 59/38 of 2 December 2004 (hereinafter the “Convention” or “UNCSI”), is the first comprehensive multilateral instrument to regulate the jurisdictional immunity of States before the courts of foreign States. Although the Convention has not formally entered into force – it requires thirty ratifications, and twenty-five States have ratified or acceded to it so far – it exerts a considerable influence on the formation of customary international law, since it expresses the opinio iuris of States and reflects their practice in the field of jurisdictional immunity. The Convention thus constitutes an important point of reference both for domestic courts and for international tribunals. Many States also regard the Convention as binding to the extent that it codifies customary law already existing in this area.

II. The commercial activity exception – Article 10 UNCSI

1. Normative basis and conditions of application

Article 10(1) UNCSI provides that if a State engages in a commercial transaction with a foreign natural or juridical person and, by virtue of the applicable rules of private international law, differences relating to that transaction fall within the jurisdiction of a court of another State, the State cannot invoke immunity from jurisdiction in a proceeding arising out of that transaction. The provision thus lays down four basic conditions:

(i) the participation of the State in the transaction;

(ii) a relationship between the subject matter of the dispute and the transaction;

(iii) the commercial character of the transaction; and

(iv) the jurisdiction of the foreign court under the applicable conflict-of-laws rules.

It is the requirement of a commercial transaction that causes lawyers the greatest argumentative difficulty. The definition of a “commercial transaction” in Article 2(1)(c) UNCSI covers categories such as: any commercial contract or transaction for the sale of goods or supply of services; any contract for a loan or other transaction of a financial nature, including any obligation of guarantee or of indemnity in respect of any such loan or transaction; and any other contract or transaction of a commercial, industrial, trading or professional nature. The International Law Commission (hereinafter the “ILC”), in its commentary to the 1991 Draft Articles, indicated that the very notion of a “transaction” is broader in scope than that of a “contract” and also embraces activity of a non-contractual character, including commercial dealings and negotiations.[1]

It is also worth noting at this point that Article 10(2) of the Convention provides for two “exceptions to the exception”, upon whose occurrence immunity retains its force notwithstanding the commercial character of the transaction. These are two situations:

(i) where the commercial transaction is concluded between States, that is, on a government-to-government basis; and

(ii) where the parties to the transaction have expressly so agreed.

In these situations immunity remains fully operative and cannot be displaced.

2. The nature test versus the purpose test: the principal criteria for distinguishing the commercial from the sovereign character of an act under the UNCSI

The central practical issue, and the one generating the most disputes in the application of the commercial exception, is the criterion by which a given transaction is to be characterised. Put at its simplest, lawyers and judges ask themselves: how does one tell whether a given act is commercial or sovereign? To answer that question, Article 2(2) UNCSI introduces a two-stage test. According to it, in determining the commercial character of a transaction, reference should be made primarily to its nature, while its purpose should be taken into account where, in the practice of the State concerned, the purpose is relevant to determining the non-commercial character of the transaction.

The ILC commentary specifies that the tests are to be applied in sequence. The nature of the transaction is examined first. If it proves to be non-commercial or governmental at that stage, any further inquiry into purpose is unnecessary.[2] If, however, the nature points to a commercial character, the defendant State is entitled to challenge that finding by reference to the purpose of the transaction, provided that the purpose is, in its practice, relevant to characterising the transaction as non-commercial. The scheme may be illustrated as follows: the court first asks “what is this transaction in itself?”, and only where the answer suggests commercial activity does it ask “for what purpose was it concluded?”. The burden of proving the sovereign purpose of the activity rests on the State that invokes it.

3. The nature test and the purpose test in the light of State practice

State practice in the application of these analytical tools is not uniform. Although a strong emphasis on the nature test as the principal, decisive criterion is discernible,[3] there is no shortage of cases in which courts say expressly that the entire context of the transaction must be examined in order to determine its character.[4] Two main interpretative positions have accordingly emerged among States.

The first assumes a rigorous stance, manifested in the application of the “nature test” alone, regardless of the purpose of the activity. On this approach, for example, a commercial contract for the supply of boots and uniforms for the army will be characterised as economic activity, and breaches arising from it – provided they fall within the jurisdiction of a foreign court – will be adjudicated there, irrespective of the sovereign purpose of the transaction, namely the provisioning of the armed forces.[5] The very nature of a purchase of uniforms from a private supplier is commercial, and in that relationship the State behaved like a private party. The State’s motivation, by contrast, is irrelevant to the assessment of the transaction.

The second position adopts a somewhat more flexible approach, taking into account, in assessing the transaction, its purpose and the situational context of its conclusion (including, where contracts concluded by State-owned enterprises are concerned, the level of State control and the ownership structure). This conception is also endorsed by the text of the UNCSI, which permits an analysis of purpose where it is relevant to determining non-commercial character.

4. Case law – how courts apply the nature and purpose tests in practice

(a) The ICJ in Certain Iranian Assets (2023)

In Certain Iranian Assets (Islamic Republic of Iran v United States of America) (Judgment of the ICJ of 30 March 2023), the International Court of Justice, addressing the question of jurisdiction over claims concerning Bank Markazi – the central bank of Iran – applied reasoning functionally convergent with the nature-and-purpose test, even though the case did not formally concern the jurisdictional immunity of the State but rather the interpretation of the notion of a “company” within the meaning of the 1955 Treaty of Amity. The Court made clear that the question of “company” status and the rules on sovereign immunity constitute two distinct normative orders, and then held that an entity performing exclusively sovereign functions cannot be characterised as a “company” within the meaning of the Treaty, whereas an entity combining commercial activity with sovereign functions may acquire that status only to the extent that it actually carries on commercial activity “alongside” its sovereign functions. A key element of the Court’s reasoning was its rejection of an approach that examines the character of a transaction “as such”, in isolation from its context – the Court stressed that a transaction must be placed in its context, having regard to its links with the exercise of a sovereign function, and found that the purchase and management of securities by Bank Markazi was merely a way of exercising its essential sovereign function of managing the State’s currency reserves, and not commercial activity carried on independently of those functions. Although this reasoning was formulated on a treaty basis rather than in the context of the commercial exception in Article 10 of the 2004 UN Convention, its logic is structurally identical to the moderate approach to the nature-and-purpose test embodied in the Convention: the nature of a transaction cannot be assessed in the abstract but requires an examination of its functional embedding in the entirety of the entity’s activity, and in particular of its connection with the exercise of sovereign prerogatives – an argument against a mechanical application of the nature criterion in isolation from the institutional and functional context of the act in question. The Court found in that case that acts undertaken “within the framework and for the purposes of” a sovereign objective do not constitute commercial activity but a means of carrying out sovereign functions.[6]

(b) Sovereign bond issues – the Argentine cases under the FSIA

Republic of Argentina v Weltover, Inc (504 US 607 (1992)) is one of the foundational decisions of the Supreme Court of the United States on the application of the commercial exception under the Foreign Sovereign Immunities Act of 1976 (FSIA), and thus bears directly on the displacement of the jurisdictional immunity of the State. The facts concerned Argentine bonds known as “Bonods”, which the Republic of Argentina issued in the early 1980s as part of a governmental currency stabilisation programme: unable to cover its obligations under foreign exchange insurance contracts (FEICs), Argentina issued bonds denominated in US dollars, payable at the creditor’s election on the markets in London, Frankfurt, Zurich or New York. When in 1986 Argentina unilaterally extended the payment deadlines and offered creditors substitute instruments in place of redemption of the overdue bonds, two Panamanian creditors and a Swiss bank – refusing the restructuring and designating New York as the place of performance – brought an action for breach of contract before the federal court in New York. Argentina contested the court’s jurisdiction, invoking jurisdictional immunity.

The Supreme Court, deciding the case unanimously through the pen of Justice Scalia, applied the nature test flowing directly from § 1603(d) FSIA, which expressly requires the commercial character of an activity to be determined by reference to its nature rather than to the purpose for which the State undertook it. The Court squarely rejected Argentina’s argument that the context of the bond issue – serving the sovereign objective of currency stabilisation and management of the foreign exchange deficit – should be decisive of the sovereign character of the transaction, stressing that the statute unambiguously moves away from assessing commerciality through the prism of the motive for the act. In doing so, the Court formulated the key criterion of distinction: a State acts “commercially” where it does not perform the role of a regulator of the market but behaves like a private player within the market, undertaking acts which private parties could perform in the same way.[7] The Bonods were held to be instruments corresponding in every respect to standard debt instruments of private commerce: they were negotiable, dollar-denominated, capable of being held by private parties, and provided a stream of monetary income – none of these features exhibited characteristics reserved exclusively for sovereign activity. The Court further rejected the proposition that an analysis of the entire context of the transaction was necessary in order to establish its commerciality, observing laconically that even viewed in their full context the bonds contained nothing that was not analogous to a private commercial transaction – except precisely their purpose, which by force of the statute remains irrelevant. It is worth noting that the approach adopted in Weltover represents the rigorous variant of the nature test, close to the Anglo-American approach: the purpose of the State’s act is excluded from the analysis altogether, which contrasts both with the moderate position adopted by the UNCSI (which allows purpose to be taken into account as an auxiliary criterion) and with the functional, contextual approach applied by the ICJ in Certain Iranian Assets (2023).

(c) The foreign activity of a State-owned company – Beijing Urban Construction Group v Republic of Yemen (ICSID, 2017)

Beijing Urban Construction Group Co Ltd v Republic of Yemen (ICSID Case No ARB/14/30, Decision on Jurisdiction, 31 May 2017) admittedly concerned the jurisdiction ratione personae of an arbitral tribunal in an investment dispute under the China–Yemen bilateral investment treaty, rather than the jurisdictional immunity of the State in the strict sense; nevertheless, the tribunal’s reasoning on the characterisation of the activity of a State-owned enterprise displays significant structural affinities with the nature test applied in the context of the commercial exception. The facts concerned the Chinese State-owned enterprise BUCG which, having won an international tender, concluded a contract in 2006 for the construction of a new terminal at Sana’a airport. In 2009 Yemen was alleged to have used military force to expel BUCG from the construction site and to have expropriated its investment. Yemen contested the tribunal’s jurisdiction, arguing that BUCG, as a wholly State-owned company, was an emanation of the Chinese State and therefore lacked the status of a “national of another Contracting State” within the meaning of Article 25(1) of the ICSID Convention.

The tribunal, rejecting that objection, applied the so-called Broches test, under which a State-owned enterprise is excluded from the personal scope of the Convention only where it acts as an agent of the government or performs essentially governmental functions. What matters for present purposes is that the tribunal clearly separated the purpose of the activity from its nature, stressing that the test must be applied to the specific transaction in dispute, not to the general ownership structure or the strategic objectives of the enterprise. Although BUCG operated in line with general State policy guidance, and the performance of the contract might serve governmental objectives, the tribunal held that the construction activity itself – performed under a contract concluded as the result of an open tender, on market terms and in direct competition with other contractors – was of an “essentially commercial, rather than governmental” character. This reasoning is functionally identical to the logic of the nature test applied within the commercial exception to immunity: the purpose of the entity’s activity (the implementation of State policy) is irrelevant to the legal characterisation of the act if its external form and the manner of its performance correspond to the model of activity of a private party – in this case, a general contractor operating on market terms. The case thus supplies an argument for the proposition that the distinction between the nature and the purpose of an act is a principle of a general character, extending beyond the framework of State immunity and present in the wider system of international investment law.

5. How does a court assess in practice whether an activity is commercial? The component elements of the nature test

In the judicial and doctrinal practice of States, a series of criteria has crystallised which assists lawyers in properly characterising the activity of a State. These criteria do not form an exhaustive catalogue or a rigid hierarchy – they are applied jointly and contextually.

The “private counterpart” criterion (private party test)

The fundamental question is: could the act in question have been undertaken by a private party? If the answer is affirmative, the activity is commercial in character. The purchase of office supplies by a ministry, the conclusion of a lease of a building for an embassy, the issue of bonds on the financial market – in all these cases there exists a private counterpart in the form of a business or an investor performing analogous acts. If, by contrast, the act by its very essence can be performed only by a sovereign – for example, the grant of a licence for the extraction of mineral resources, the establishment of an economic zone, or the imposition of an embargo – then no private counterpart exists and the act must be characterised as acta iure imperii.

The market participant criterion (market participant test)

This criterion is related to the one described above, but places the emphasis on the manner in which the State behaves. The questions asked here are the following: does the State enter the market on equal terms, like a private party? Does the State act on the market in a manner characteristic of private actors? Does the State refrain from taking advantage, on the market, of any privileges characteristic of a sovereign? In Republic of Argentina v Weltover, Inc (504 US 607 (1992)), already mentioned above, the US Supreme Court held that an act is commercial in character where the State “does not act as a market regulator, but in the manner of a private player within the market”.[8] A good example of commercial activity in the light of the market participant criterion is the situation analysed by the ICSID tribunal in Beijing Urban Construction Group v Republic of Yemen. The arbitral tribunal held that participating in a tender on an equal footing with other private bidders and negotiating the terms of the contract are behaviours typical of a market participant.[9] Where, by contrast, the State makes use of instruments of public authority unavailable to private parties (statute, decree, concession), it cannot then be treated as a market participant.

The profit motive criterion

The ILC commentary indicates that, according to one of the conceptions proposed in the course of the work on the Convention, the profit motive constitutes the most important criterion of the commerciality of a transaction.[10] In practice, however, the criterion is merely auxiliary in character: its presence reinforces a characterisation as acta iure gestionis, but its absence does not automatically determine the sovereign character of the act. A profit motive is, by its very essence, what we associate with the conduct of economic and commercial activity. It must be remembered, however, that many State transactions – such as the purchase of office supplies for an embassy, or agreements signed with scientific and research institutions covering, for example, State funding of research – are commercial in nature even though their immediate objective is not pure profit in the economic sense of the word.

The regulatory criterion

Of fundamental importance is the distinction between regulatory acts (the issue of licences, administrative decisions and permits, the application of customs provisions, the imposition of sanctions) and contractual acts based on a genuine contractual relationship in which the parties have equal rights and neither of them wields any attribute of public authority. Acts in the regulatory sphere are an attribute of sovereignty and are invariably classified as iure imperii. This is so because the State then occupies a privileged position, having a greater or exclusive influence on the shape of the legal relationship, but also because, as a regulator, the State decides – in the name and on behalf of its citizens – about their common goods, for example natural resource deposits. The example of resource extraction also perfectly illustrates the interconnection between the various criteria. Under the criterion here described, when a government grants a concession for the extraction of crude oil, it acts as a regulator. When, on the other hand, it concludes a contract with a contractor for drilling in the territory of another country, this may be characterised as the activity of a contracting party which occupies a position equal to that of the contractor in the transaction. Pursuing the mining industry example, let us then imagine that a State grants a concession to a private company for the extraction of resources in its territory. Knowing the market regulator criterion, we will conclude that it acts in the sphere of iure imperii, since by its act it conditions access to a good belonging to all its citizens – something only a sovereign endowed by the Nation with the appropriate competences can do. In the same situation we may also ask whether a private party could have done this, or whether in this transaction the State is taking advantage of some special privileges giving it a better position. Is the State behaving like a private player on the market? As this example shows, all the criteria listed above interlock and complement one another.

The genuine contractual relationship criterion

The commercial exception requires the dispute to arise out of a commercial transaction. The commercial character of an act may therefore also be indicated by the existence of a real contractual relationship within which the State has assumed an obligation similar to that of a private person. Simplifying, what is at issue is the existence of a contract binding on the State whose clauses, subject matter and guarantee mechanisms show that the parties acted as equal participants in commerce. Where the dispute arises precisely out of a contract so shaped, immunity cannot serve as a shield against contractual liability.

6. Acts not constituting commercial activity

For the sake of completeness, it is worth identifying the categories of activity which – irrespective of their external form – remain within the sphere of iure imperii and are not subject to the commercial exception. There is no coherent and binding catalogue in this respect, but it is emphasised in the literature that such acts include, above all: the issue of licences and permits; the imposition of taxes and customs duties; the enactment of legislation and implementing acts; the conduct of foreign and defence policy (including in the field of economic security); the conclusion of intergovernmental agreements creating public-law institutions and funds; and the issue of administrative decisions restricting or rationing access to the market. These acts are regulatory and sovereign in character; their essence is the exercise of authority to which a private party has no access.

7. Typical interpretive errors

In legal practice, the most serious error is to conflate the purpose of an act with its nature. For example, a court, or counsel for the State party, may argue that the purchase of soldiers’ boots for a military unit serves the sovereign functions of national defence and is therefore not subject to the commercial exception, its nature being sovereign. But is that really so? In such a situation one should return to the auxiliary criteria of the nature test set out above. Could a private person have made such a purchase? Essentially yes: there are no prohibitive regulations, nor any special entitlements a person must hold, in order to purchase soldiers’ boots, even in large quantities. Did the State act like a private party in this relationship? Yes: the State concluded a contract of sale on terms equal to those of other market participants (we assume, for the purposes of the hypothetical, that it enjoyed no special privileges in this respect). And finally, is there a contract whose content points to a commercial character? A contract of sale is without doubt commercial in nature, being the basic tool of economic exchange. The nature of the transaction under discussion is therefore commercial. Once the nature has been determined, it becomes easier to separate it from the purpose, which in this situation was the provisioning of the army and, in the broader context, no doubt action in the interest of a sound state of national defence. The further characterisation of that purpose, and the determination of how far it matters in the specific factual situation for the characterisation of the transaction, is for the court.

Another frequent error is the automatic attribution of a sovereign character to the activity of companies owned by the State. As follows from the ILC Draft Articles, State enterprises are covered by a presumption that they are not entitled to perform governmental functions.[11] Under Article 10(3) UNCSI, where a State enterprise or other entity established by a State has independent legal personality and is capable of suing or being sued and of acquiring, owning and disposing of property, including property which the State has authorised it to operate or manage, and it is involved in a proceeding relating to a commercial transaction in which it is engaged, the immunity from jurisdiction enjoyed by the State itself remains unaffected. Such a company contracts obligations on its own account, while State immunity remains intact; no “piercing of the corporate veil” occurs. Only if it is proved that a particular State-owned company in fact exercised prerogatives of sovereignty can one speak of its acquiring the status of an “instrumentality of the State” and of the State’s immunity extending to that company as well. The burden of proof, however, rests on the company, or on the State invoking immunity.

A third error consists in confusing the scope of the notion of a “transaction” with that of a “contract”. As already indicated, a transaction also embraces dealings and pre-contractual conduct. This means that claims derived from the negotiation stage may fall within the commercial exception even if no contract was ultimately concluded.

III. Conclusion

The restrictive doctrine of State immunity is an important element of contemporary public international law, although the text of the Convention itself is in need of “refreshing”, particularly as regards laying down clear criteria for the characterisation of State conduct, consistent with the general practice of States. Is that practice, however, currently uniform enough to justify a further attempt at codification? That question remains, unfortunately, difficult to answer, since States are as a rule reluctant to agree to any limitations of their sovereignty – above all to limitations which, in the economic sphere, could entail the necessity of submitting to the jurisdiction of a foreign State. It is not inconceivable, however, that the dynamic development of business and advancing globalisation will compel countries to accept certain restrictions in the name of mutual trust in economic transactions and the harmonious development of the market.

The commercial exception (Article 10 UNCSI), although seemingly simple in conception, gives rise to interpretive difficulties in practice. The keys to its interpretation are: avoiding the automatic transposition of the purpose of an act onto the assessment of its nature; a harmonious weighing of the criteria in assessing the character of the State’s activity, so as to arrive at the result closest to the truth; and refraining from automatic determinations concerning State-owned companies based solely on their formal structure, in isolation from the acts the company actually performs. The case law, from Weltover to Certain Iranian Assets, provides the lawyer with solid tools for conducting such an analysis. The practitioner should ask, in turn: could this act have been performed by a private party? Did the State behave like a market participant? Is there a genuine contractual relationship? Only a negative answer to these questions opens the way to considering the sovereign character of the act. In the author’s view, the drafters of the Convention did a genuinely good job in codifying the commercial exception, taking an intermediate position which also allows the purpose of a transaction to be taken into account in its analysis.

Jurisdictional immunity is not an unlimited privilege of the sovereign. And although the proponents of the principle par in parem non habet imperium – as an expression of the sovereignty of States, which is the foundation of the international order – oppose the codification of exceptions to immunity, especially in the context of State-owned enterprises, it must be remembered that the world does not stand still, and States engage in far more extensive economic activity than they did a few decades ago. It is in the interest of States and private-sector actors alike to provide judicial authorities with the tools that will enable them to resolve disputes arising out of commercial transactions and to decide before which court a given State will answer. In today’s business realities, the sovereign must reckon with accountability before the courts of the States of the parties with which it deals. That is the price of the global activity of States in the twenty-first century.

Bibliography

I. Legislation and treaties

Foreign Sovereign Immunities Act of 1976, 28 USC §§ 1603(d), 1605 [USA].

Regulation (EU) No 1215/2012 of the European Parliament and of the Council of 12 December 2012 on jurisdiction and the recognition and enforcement of judgments in civil and commercial matters (Brussels I bis).

State Immunity Act 1978 [United Kingdom].

State Immunity Act, RSC 1985, c S-18 [Canada].

Treaty of Amity, Economic Relations, and Consular Rights between the United States of America and Iran, Tehran, 15 August 1955.

United Nations Convention on Jurisdictional Immunities of States and Their Property, adopted by UN General Assembly resolution 59/38 of 2 December 2004 [UNCSI], arts 2, 10.

II. Case law

Beijing Urban Construction Group Co Ltd v Republic of Yemen, ICSID Case No ARB/14/30, Decision on Jurisdiction, 31 May 2017.

Certain Iranian Assets (Islamic Republic of Iran v United States of America) (Judgment of 30 March 2023).

I Congreso del Partido [1983] 1 AC 244 (HL) (Lord Wilberforce).

Jurisdictional Immunities of the State (Germany v Italy: Greece intervening) (Judgment) [2012] ICJ Rep 99.

Republic of Argentina v Weltover, Inc 504 US 607 (1992) [Supreme Court of the United States].

III. Work of the International Law Commission

Draft Articles on Jurisdictional Immunities of States and Their Property, with Commentaries, Yearbook of the International Law Commission 1991, vol II, pt 2 (UN 1994), Commentary to art 2, paras 15, 20, 25.

UN Doc A/C.6/45/SR.25.

UN Doc A/C.6/59/SR.13.

IV. Literature

Fox H and Webb P, The Law of State Immunity (revised and updated 3rd edn, Oxford University Press 2015).

Thomas KR, The Commercial Activity Exception to State Immunity: An Introduction (Edward Elgar 2024).

[1]Draft Articles on Jurisdictional Immunities of States and Their Property, with Commentaries, Yearbook of the International Law Commission 1991, vol II, pt 2 (UN 1994), Commentary to art 2, para 20.

[2]Draft Articles on Jurisdictional Immunities of States and Their Property, with Commentaries, Yearbook of the International Law Commission 1991, vol II, pt 2 (UN 1994), Commentary to art 2, para 25.

[3]Beijing Urban Construction Group Co Ltd v Republic of Yemen, ICSID Case No ARB/14/30, Decision on Jurisdiction, 31 May 2017, [35].

[4]I Congreso del Partido [1983] 1 AC 244, 267 (Lord Wilberforce).

[5]KR Thomas, The Commercial Activity Exception to State Immunity: An Introduction (Edward Elgar 2024) 12.

[6]Certain Iranian Assets (Islamic Republic of Iran v United States of America) (Judgment of 30 March 2023) [50].

[7]Republic of Argentina v Weltover, Inc 504 US 607, 614 (1992).

[8]Weltover (n 7) 614.

[9]Beijing Urban Construction (n 3) [40].

[10]ILC Draft Articles (n 1), Commentary to art 2, para 20.

[11]ILC Draft Articles (n 1), Commentary to art 2, para 15.

Artykuł The Commercial Activity Exception pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

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.

Introduction

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.

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[1] establishes rigorous procedures and imposes specific restrictions on providers, deployers and other links in the artificial intelligence value chain. In relation to the GDPR[2], 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.

1st         Who Is Who in the AI Ecosystem: Provider, Deployer, Controller

a)    Provider or Deployer? The Answer Determines the Scope of Liability

Defining an organisation’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’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.

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 (output). 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.

b)    When an AI System Becomes a High-Risk System

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).

c)     Transparency, Instructions for Use and Human Oversight: Obligations in Practice

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’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.

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.

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, “human oversight shall aim to prevent or minimise the risks to health, safety or fundamental rights (…)”. 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, “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”.

d)    Training and Inference: Where the Data Risk Hides in Language Models

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’s operation: training and inference.

Training phase

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.

Inference phase

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’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’ and customers’ 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.

2nd    Bias, Hidden Scoring and Explainability: Algorithmic Risks under Article 5 GDPR and Article 10 AI Act

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.

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, “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”, 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’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’ operation, adequate human oversight and the ability to verify and challenge the results generated by AI.

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’s operation.

3rd      Can an Employer Require the Use of AI? The Limits of Lawfulness in the Workplace

Assessing the lawfulness of deploying artificial intelligence systems in the workplace requires a precise delineation of the boundary between the employer’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’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’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.

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’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’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’s lack of genuine freedom of choice in the relationship with the employer makes such consent extremely difficult to defend before the supervisory authorities.

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 social scoring (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.

4th       Individuals versus the Algorithm: Erasure, Human Intervention, Transparency

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’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’s right to demand the erasure of their data and the architecture of modern technologies.

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 (human-in-the-loop), 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’ 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.

5th       Four Risk Tiers and a New Supervisory Map: KRiBSI alongside the President of the Personal Data Protection Office

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.

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’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 (privacy and compliance by design) exposes controllers to a serious risk of financial penalties and administrative sanctions in the form of an order to shut down the algorithm immediately.

6th       DPIA and FRIA: from a Document in a Drawer to a Dialogue with the Regulator

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.

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’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.

7th       AI Governance: How to Implement AI in an Organisation and Be Able to Prove It

Effective management of artificial intelligence systems within an organisation requires a multi-level oversight model (AI Governance[3]) 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 adversarial attacks[4], attempts to poison training data sets (data poisoning) and model inversion techniques aimed at extracting the personal data used to build the model.

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.

Summary and Conclusions

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’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.

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’ rights against the algorithm requires advanced procedures such as machine unlearning, and the unconditional guarantee of real rather than merely token human oversight (human-in-the-loop).

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.

Consequently, the lawful and ethical use of artificial intelligence in modern business is not possible without a comprehensive oversight model: AI Governance. Traditional IT security measures must be extended to include mechanisms that harden models against a new generation of cyberattacks, such as adversarial attacks and data poisoning. The synergy of the principles of privacy by design and privacy by default 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.

Bibliography

Legislation

1. 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
2. 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
3. Act of 26 June 1974 – Labour Code (consolidated text: Journal of Laws of 2025, item 277, as amended).

Online Sources

• Letter of [date] issued by the President of the Personal Data Protection Office, ref. DOL.401.354.2024, Comments of the PUODO on the draft Act on Artificial Intelligence Systems, www.uodo.gov.pl.
• Wolters Kluwer: “Przetwarzanie danych osobowych na podstawie prawnie uzasadnionego interesu administratora” [Processing of personal data on the basis of the controller’s legitimate interest], 12.12.2025, available online: https://www.lex.pl/test-rownowagi-rodo,43864.html
• European Parliamentary Research Service (EPRS): Artificial Intelligence Act: State of play and implementation challenges, European Parliamentary Research Service (EPRS) At a Glance Briefing, 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).
• Biznes Myśli: Fine-tuning LLM – fakty i mity. Jak skutecznie dotrenować duży model językowy [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).
• LegalGeek: AI Act vs RODO. Jak sztuczna inteligencja wpływa na ochronę danych osobowych [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).
• Wikipedia (the free encyclopedia): Duży model językowy [Large language model], [online], available at: https://pl.wikipedia.org/wiki/Du%C5%BCy_model_j%C4%99zykowy (accessed: 29.05.2026).
• 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).

[1] 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

[2] 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

[3] 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).

[4] adversarial attacks – the deliberate manipulation of input data in order to deceive the algorithm.

Artykuł A Human in the Loop, Data under Control. The GDPR and the AI Act in Implementation Practice pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

KG LEGAL Kiełtyka Gładkowski was pleased to be represented at the ABA International Law Section Fall Conference, held in London from 21–25 September 2026 at the magnificent Mansion House.

The conference brought together an outstanding international community of legal professionals and provided a valuable forum for exchanging knowledge, sharing practical experience and discussing the latest developments affecting cross-border legal practice. The programme covered a broad range of topics, including international disputes, the use of AI in legal proceedings, digital assets and asset recovery, arbitration enforcement, corporate transactions, privacy and regulatory challenges.

Our partner Małgorzata Kiełtyka participated as a speaker in the panel “Heed Caution When Using Contractual Terms Such as ‘Best Knowledge’, ‘Best Efforts’ or ‘Material Adverse Change’ Outside the United States”.

We would like to thank the moderator, Randall Hanson, and fellow speakers Jörg Rehder, Alberto Predieri and Mark Howard for an engaging and highly practical discussion.

For reference, we have also included a PDF containing the detailed programme and panel information from the conference.

The conference provided an excellent opportunity to exchange views with colleagues from different jurisdictions and to strengthen professional connections within the international legal community. The knowledge and perspectives shared during such events contribute directly to our day-to-day work and help us develop practical solutions for clients involved in complex cross-border matters.

We are also pleased to share a short video recorded outside Mansion House, in which our colleague reflects on the conference and expresses our appreciation to the American Bar Association, the International Law Section, Section Chair Ed Mullins and Angela Benson for their hospitality and excellent organisation.

We thank all colleagues, speakers and friends who contributed to making the London conference a valuable and inspiring professional experience.

Artykuł ABA Fall Conference London: Our Speaker at Mansion House pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

ABA ILS, Life Sciences and Healthcare Committee, September 2026Prepared by KIELTYKA GLADKOWSKI KG LEGAL

I. Introduction

The longevity economy is commonly estimated at several hundred billion dollars worldwide and continues to grow rapidly.[1] Venture capital flows into biotechnology companies that promise to slow, halt, or reverse aging. Clinics from Zurich to Seoul to Miami sell “longevity programmes.” Supplement brands attach the word “longevity” to products ranging from NAD precursors to collagen powders. And yet the word “longevity” does not exist as a legal category in European Union law. There is no definition, no register, no certification scheme, and no authority empowered to decide what may or may not describe itself as a longevity product, a longevity clinic, or a longevity company.

The gap between an enormous market and a non-existent legal definition is precisely where legal risk accumulates. Where there is no definition, there is classification risk, claims risk, and enforcement risk. There is also a phenomenon that this article, borrowing from the vocabulary of environmental marketing, calls “longevity washing.”

This article examines the longevity industry through the lens of EU regulatory law. The EU framework governs a market of some 450 million consumers and, because it is horizontal, sector-specific, and increasingly enforcement-oriented, it offers a useful point of comparison for practitioners in the United States, Asia, and Latin America. Part II explains why “longevity” is not a legal category and introduces the borderline-product problem that follows from that fact. Part III develops the concept of longevity washing and its consumer-facing, unfair-commercial-practice, and investor-facing dimensions. Part IV addresses the regulation of artificial intelligence in longevity research and diagnostics, including the revised timeline of the AI Act, the position of the European Medicines Agency on AI in the medicinal-product lifecycle, and the data-protection architecture formed by the GDPR and the European Health Data Space. Part V maps the sectoral “hot spots” across pharmaceuticals, medical devices, cosmetics, and food supplements. Part VI draws conclusions for practitioners.

II. “Longevity” in the Eyes of the Law: The Borderline-Product Problem

In EU regulatory law, aging is not a disease. The European Medicines Agency has never approved a medicinal product with “aging” as a therapeutic indication, and under Directive 2001/83/EC a medicinal product is defined by reference to the treatment or prevention of disease or the restoration, correction, or modification of physiological functions.[2] The position in the United States is comparable. The long-running debate around the Targeting Aging with Metformin (TAME) trial illustrates the point: its designers had to negotiate a trial protocol with the Food and Drug Administration that targets aging indirectly, through a composite of age-related diseases, precisely because “aging” itself is not an approvable indication.[3]

The consequence is profound. A company cannot lawfully market a product “to treat aging.” The entire longevity industry therefore operates in the space of indirect claims: “supports healthy aging,” “promotes cellular health,” “extends healthspan.” Each of those phrases pushes the product toward one of several very different regulatory regimes.

This is the borderline-product problem, and it is the single most important legal concept in the sector. The same molecule – an NAD precursor, for example – may be: a medicinal product, if it is presented as treating or preventing disease or exerts a significant pharmacological, immunological, or metabolic effect; a food supplement under Directive 2002/46/EC;[4] a novel food requiring pre-market authorisation under Regulation (EU) 2015/2283;[5] a cosmetic under Regulation (EC) No 1223/2009, if applied to the skin with a claim limited to appearance;[6] or a medical device. The last category is broader than it may appear: since the Medical Device Regulation (MDR) came into application, products listed in its Annex XVI – dermal fillers, certain energy-based aesthetic equipment – are regulated as devices even where they have no medical purpose at all.[7]

Classification does not follow the label a company chooses. It follows presentation, composition, and route of administration. The Court of Justice of the European Union has held for four decades that a product is a medicinal product “by presentation” where it is described or recommended as having properties for treating or preventing disease, irrespective of whether it actually has those properties; and that classification “by function” turns on a case-by-case scientific assessment of the product’s actual pharmacological, immunological, or metabolic action, not on the manufacturer’s intention.[8] Marketing language alone can therefore reclassify a product and, with it, the entire compliance burden: marketing authorisation, good manufacturing practice, pharmacovigilance, and advertising restrictions.

When a client says “we are a longevity company,” the first question counsel should ask is: which regime are you actually in? “Longevity” is a marketing word. The law knows only medicines, devices, foods, and cosmetics.

III. Longevity Washing: When the Claim Outruns the Science

Greenwashing – claiming environmental virtue one does not have – is a familiar concept. Longevity washing is its younger sibling: labelling a product, a clinic, a research programme, or an entire company as “longevity-focused” when the science, the indication, or the business model does not support that description. The practice matters legally for three reasons.

A. Consumer-Facing Claims

Under the Health Claims Regulation, any health claim made on a food – and food supplements are foods – must be specifically authorised and listed in the EU register.[9] Claims such as “slows aging,” “supports cellular repair,” or “extends lifespan” are not authorised health claims. National authorities in the EU and the United Kingdom have treated phrases such as “cellular repair” and even “healthy aging” as impermissible where they imply a physiological mechanism or a reduction of disease risk. A substantial part of the longevity supplement market is thus operating with claims that would not survive enforcement scrutiny.

B. Unfair Commercial Practices and the Greenwashing Template

The second layer is the general prohibition of misleading commercial practices under the Unfair Commercial Practices Directive (UCPD).[10] Here the EU has recently built an entire enforcement architecture against greenwashing that repays close study by anyone advising on health claims. The Empowering Consumers for the Green Transition Directive amends the UCPD to prohibit generic environmental claims – “eco-friendly,” “climate-neutral” – unless the trader can demonstrate recognised excellent environmental performance, and it adds a series of specific greenwashing practices to the UCPD’s blacklist. Its provisions apply across the Union from 27 September 2026.[11]

The Directive should be read as a template. If “climate-neutral” now requires substantiation, it is a very short doctrinal step for regulators and courts to demand the same of “clinically proven to slow biological aging.” The methodology of anti-greenwashing enforcement – generic claim, absence of substantiation, misleading omission – maps directly onto longevity claims. The UCPD’s general clauses on misleading actions and omissions already provide the legal basis; what the green-claims reform supplies is a worked example of how those clauses can be operationalised against vague virtue-signalling.

C. Investor-Facing Washing

The third dimension is of particular relevance to transactional lawyers. Start-ups increasingly brand themselves as “longevity biotech” because that is where the capital is, even when the underlying asset is a conventional dermatology product or a wellness application. For investors, this creates due-diligence risk: is the “longevity platform” in fact a regulated medicinal pipeline, with all the timeline and cost that implies, or a supplement business one enforcement letter away from relabelling? Misclassification discovered after investment is a valuation event. For the company, aggressive longevity framing in fundraising materials can generate liability under securities and misrepresentation rules in multiple jurisdictions. In due diligence on a longevity start-up, the regulatory classification memorandum is not an annex to the deal; it is the deal.

IV. Artificial Intelligence in Longevity Research and Diagnostics

The longevity field is arguably the most AI-dependent sector in the life sciences, for a simple reason: aging is not one target but thousands of interacting biological processes. A geroprotective drug cannot be designed the way an antibiotic is designed. The entire aging phenotype must be modelled, and that means machine learning on large multi-omic datasets. AI is not an add-on in longevity; it is the core methodology. That raises three distinct regulatory questions under EU law.

A. The AI Act

Regulation (EU) 2024/1689, the AI Act, is the most comprehensive horizontal AI regulation in force anywhere and applies to almost every AI system used in the longevity value chain.[12] It adopts a risk-based structure – prohibited practices, high-risk systems, transparency obligations, and minimal-risk systems – and the critical category for this sector is high-risk. An AI system that is itself a medical device, or a safety component of one, under the MDR or the In Vitro Diagnostic Regulation, and that is subject to third-party conformity assessment under those instruments, is high-risk under Article 6(1) and Annex I of the AI Act.[13] This captures, for example, an AI tool that estimates a patient’s “biological age” from a blood panel and uses that estimate to guide a clinical intervention; an AI-driven diagnostic that predicts age-related disease risk; and the software layer of many longevity-clinic platforms once it makes a diagnostic or prognostic claim.

High-risk status is not a prohibition; it is a compliance regime. It requires a risk-management system, data governance with representative and bias-tested training data, technical documentation, logging, human oversight, transparency to deployers, accuracy, robustness and cybersecurity, a conformity assessment, and post-market monitoring.[14] For most longevity start-ups this is a wholly new compliance layer on top of MDR conformity assessment, and the two must be coordinated.

The timeline has recently changed, and practitioners should update their advice accordingly. The original AI Act applied the high-risk obligations from 2 August 2026 for stand-alone systems listed in Annex III and from 2 August 2027 for AI embedded in products covered by Annex I, including medical devices. The Digital Omnibus on AI, adopted as Regulation (EU) 2026/1744 and in force since 27 July 2026, defers those dates to 2 December 2027 and 2 August 2028 respectively.[15] The deferral does not lighten the substantive regime; the prohibitions in Article 5, the obligations of providers of general-purpose AI models, and the transparency obligations of Article 50 (which apply from 2 August 2026) are unaffected. For longevity companies whose AI is a medical device, the practical effect is a compliance window of roughly two years, during which harmonised standards and notified-body capacity are expected to mature. The window is an opportunity to design AI Act compliance into the MDR technical file from the outset rather than to defer the exercise.

B. AI in Drug Discovery and the European Medicines Agency

The position for drug discovery is more nuanced. An AI model used purely internally to identify candidate molecules – the classic “AI for drug discovery” use case that dominates longevity-biotech pitch decks – is generally not, in itself, a high-risk AI system under the AI Act, because it does not make or materially influence decisions about individual patients. The regulatory question shifts instead to the European Medicines Agency. The EMA’s Reflection Paper on the use of AI in the medicinal product lifecycle sets out the Agency’s expectations from discovery through pharmacovigilance.[16] Its core message is that the burden of proof rests on the sponsor to demonstrate that the AI model is fit for its regulatory purpose. That entails documented data provenance, model validation, a degree of explainability appropriate to the stage of development, and lifecycle management of the model itself.

If a longevity company tells its investors that “our AI discovered this molecule,” the regulator will eventually ask to see the model, the training data, and the validation. Vague AI claims in fundraising materials do not survive regulatory scrutiny and, increasingly, do not survive investor due diligence either.

C. Data: The GDPR and the European Health Data Space

The third question concerns data, and it is here that longevity becomes uniquely difficult. Aging research is inherently longitudinal and multi-omic: genomic, epigenetic, and proteomic data, wearable data, and medical records, ideally across decades. Under the General Data Protection Regulation this is special-category data – health, genetic, and biometric data – and the available lawful bases are narrow.[17] The scientific-research provisions of Article 9(2)(j) and Article 89 assist, but they are implemented differently in each Member State, which produces real fragmentation for pan-European longevity cohorts.

The European Health Data Space Regulation adds a second layer.[18] It creates, for the first time, a mandatory EU-wide framework for the secondary use of electronic health data, administered by health-data-access bodies in each Member State. For longevity research the EHDS is potentially transformative, because it opens access to population-scale health data for research purposes. It also imposes strict conditions: purpose limitation, processing only within secure processing environments, a prohibition on re-identification, and specific prohibitions on using the data for advertising or marketing, for decisions detrimental to individuals such as insurance or employment decisions, or for developing products harmful to health.[19] The secondary-use chapter applies from March 2029 for most data categories, but the lead time is short in the context of longitudinal cohorts. Longevity companies that intend to monetise health data – and many do – should be mapping their data flows against the EHDS regime now.

V. Sector-by-Sector Regulatory Hot Spots

Four sectoral silos together make up the longevity industry, and the same start-up frequently straddles two or three of them without realising it. In each silo, specific “hot spots” – advertising, manufacturing standards, registration, and market withdrawal – recur in practice.

A. Pharmaceuticals

True longevity drugs – rapamycin analogues, senolytics, geroprotectors – sit squarely within Directive 2001/83/EC and its implementing rules, and will continue to do so under the new pharmaceutical package once it applies. Three practical hot spots stand out. First, off-label prescribing of rapamycin and metformin for longevity indications is widespread in private clinics across Europe. Off-label prescribing as such is a matter of clinical freedom, but off-label promotion by or on behalf of the marketing-authorisation holder is prohibited: advertising must conform to the summary of product characteristics, and advertising of unauthorised products or indications is not permitted.[20] Longevity clinics are increasingly the vehicle through which promotional messaging reaches patients, and enforcement authorities have begun to take notice. Second, good manufacturing practice: anything presented as a medicine must be manufactured under an EU manufacturing authorisation and EU GMP, with qualified persons and full traceability; compounding pharmacies producing “longevity peptides” have become a significant enforcement target. Third, unlicensed importation: a large part of the longevity-peptide market operates through personal importation and grey channels, and customs enforcement across the EU has visibly tightened.

B. Medical Devices

The MDR has fundamentally reshaped this space. Aesthetic devices – dermal fillers, energy-based devices used in longevity clinics, at-home devices marketed for “cellular rejuvenation” – are caught by Annex XVI even without a medical purpose. Three hot spots follow. Classification: many longevity wearables and applications become software as a medical device once they make a diagnostic or prognostic claim. Advertising: Article 7 of the MDR prohibits misleading claims and specifically prohibits ascribing to a device functions and properties it does not have, which operates in practice as a codified anti-washing rule for devices.[21] Registration and vigilance: the first four modules of the EUDAMED database – actor registration, UDI/device registration, notified bodies and certificates, and market surveillance – have been mandatory since 28 May 2026, with legacy devices to be registered by 28 November 2026; the vigilance module is not yet mandatory, so serious incidents and field safety corrective actions continue to be reported to national competent authorities.[22] Market withdrawals happen: a device sold as a “cellular biomarker analyser” without valid clinical evidence can be removed from the market by any national competent authority, and the MDR’s market-surveillance coordination means that one withdrawal can cascade across the Union.

C. Cosmetics and Aesthetic Products

The Cosmetics Regulation governs anti-aging creams, serums, and topical actives. Two hot spots dominate. The first is the Product Information File: the responsible person must hold a complete file including a safety assessment and substantiation of every claim. The second is the Claims Regulation, Commission Regulation (EU) No 655/2013, with its six common criteria: legal compliance, truthfulness, evidential support, honesty, fairness, and informed decision-making.[23] A claim such as “reverses skin aging at the cellular level” fails the evidential-support criterion unless a robust dossier stands behind it. Enforcement is largely national, but the Cosmetic Products Notification Portal is EU-wide, and non-compliant products are routinely notified through the Safety Gate rapid-alert system.

D. Food Supplements and Novel Foods

It is in the food and supplement space that the longevity industry is most creative and most exposed. Directive 2002/46/EC governs vitamins and minerals in supplements; everything else falls under the general food law of Regulation (EC) No 178/2002[24] and, critically, the Novel Food Regulation. Nicotinamide mononucleotide (NMN), one of the most heavily marketed longevity molecules, is the paradigmatic case. NMN is treated as a novel food in the EU, which means it cannot lawfully be placed on the market as a supplement without pre-market authorisation by the Commission following an EFSA safety assessment. Nicotinamide riboside received such an authorisation in 2020.[25] NMN has not – although the picture is now moving. In March 2026 EFSA’s NDA Panel adopted a positive opinion on the safety of β-NMN in food supplements at up to 300 mg per day for adults, and the opinion was published in May 2026.[26] An EFSA opinion is a scientific assessment, not an authorisation; the Commission must still adopt an implementing act adding NMN to the Union list, and where the applicant has requested data protection the authorisation will initially be proprietary to that applicant’s specification. Until that act is adopted, NMN products on the shelves of European longevity clinics remain unauthorised novel foods. That is not a grey area; it is a non-compliant market operating in plain sight, and it is the kind of enforcement pipeline that eventually produces high-profile withdrawal actions. Once the Health Claims Regulation is added to the picture, the conclusion is unavoidable: the food and supplement leg of the longevity industry is the most legally exposed part of the entire sector.

VI. Conclusions

The longevity industry is a genuine scientific and commercial phenomenon, and much of the underlying research is serious. But lawyers advising companies, investors, clinicians, and, increasingly, regulators in this space must see it clearly: “longevity” is not a legal category. It is a marketing frame layered on top of four very different regulatory regimes. Three practical conclusions follow for anyone whose product touches European consumers.

First, always begin with classification. The word “longevity” conveys nothing about the applicable regime. Ask what the product is, how it is presented, and what it does, and let the regulatory regime follow from those answers.

Second, treat longevity claims the way green claims are now treated. The direction of enforcement is unmistakable. The Empowering Consumers for the Green Transition Directive, applicable from 27 September 2026, gives regulators a sharpened toolkit that will, sooner or later, be pointed at health and wellness claims. Substantiation is no longer optional.

Third, integrate AI Act compliance with sectoral compliance from the outset. For any longevity venture using AI in diagnostics or patient-facing decision support, the AI Act, the MDR, and the GDPR, soon joined by the European Health Data Space, form a single interlocking compliance architecture. The deferral of the AI Act’s high-risk obligations to 2027 and 2028 is a design window, not a reprieve. Retrofitting is expensive; designing for compliance is not.

The longevity industry will keep growing, and regulators will keep catching up. The lawyers who understand both the science and the sectoral map will be the ones clients call first.

[1] Longevity economy estimates vary widely by definition; see, e.g., the periodic longevity-market reports of Bank of America Global Research and McKinsey Health Institute. The figure is offered here as an order of magnitude, not a precise valuation.

[2] Directive 2001/83/EC of the European Parliament and of the Council of 6 November 2001 on the Community Code Relating to Medicinal Products for Human Use, art. 1(2), 2001 O.J. (L 311) 67 [hereinafter Directive 2001/83]. At the time of writing, the new EU pharmaceutical package (a new directive on the Union code for medicinal products for human use and a new regulation on Union authorisation procedures) has been politically agreed and is in the final stage of formal adoption; the substantive rules discussed here are carried over into the new framework, which will apply after a transitional period of approximately two years from entry into force.

[3] See Nir Barzilai et al., Metformin as a Tool to Target Aging, 23 Cell Metabolism 1060 (2016) (describing the design of the Targeting Aging with Metformin (TAME) trial around a composite of age-related diseases).

[4] Directive 2002/46/EC of the European Parliament and of the Council of 10 June 2002 on the Approximation of the Laws of the Member States Relating to Food Supplements, 2002 O.J. (L 183) 51.

[5] Regulation (EU) 2015/2283 of the European Parliament and of the Council of 25 November 2015 on Novel Foods, 2015 O.J. (L 327) 1 [hereinafter Novel Food Regulation].

[6] Regulation (EC) No 1223/2009 of the European Parliament and of the Council of 30 November 2009 on Cosmetic Products, 2009 O.J. (L 342) 59 [hereinafter Cosmetics Regulation].

[7] Regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on Medical Devices, Annex XVI, 2017 O.J. (L 117) 1 [hereinafter MDR]; Commission Implementing Regulation (EU) 2022/2346 of 1 December 2022 Laying Down Common Specifications for the Groups of Products Without an Intended Medical Purpose Listed in Annex XVI, 2022 O.J. (L 311) 60.

[8] Case 227/82, Van Bennekom, ECLI:EU:C:1983:354 (establishing that a product is a medicinal product “by presentation” where it is described or recommended as having therapeutic or prophylactic properties, irrespective of its actual efficacy); Case C-319/05, Comm’n v. Germany, ECLI:EU:C:2007:678 (garlic capsules); Case C-140/07, Hecht-Pharma GmbH v. Staatliches Gewerbeaufsichtsamt Luneburg, ECLI:EU:C:2009:5 (classification “by function” requires a case-by-case assessment of pharmacological, immunological or metabolic properties). See also Directive 2001/83, supra note 2, art. 2(2) (in cases of doubt, the medicinal-products regime prevails).

[9] Regulation (EC) No 1924/2006 of the European Parliament and of the Council of 20 December 2006 on Nutrition and Health Claims Made on Foods, arts. 10, 13, 14, 2006 O.J. (L 404) 9 [hereinafter Health Claims Regulation]. The EU Register of nutrition and health claims is maintained by the European Commission.

[10] Directive 2005/29/EC of the European Parliament and of the Council of 11 May 2005 Concerning Unfair Business-to-Consumer Commercial Practices in the Internal Market, 2005 O.J. (L 149) 22 [hereinafter UCPD].

[11] Directive (EU) 2024/825 of the European Parliament and of the Council of 28 February 2024 Amending Directives 2005/29/EC and 2011/83/EU as Regards Empowering Consumers for the Green Transition, OJ L, 2024/825, 6.3.2024 [hereinafter ECGT Directive]. Member States were required to transpose the Directive by 27 March 2026; its provisions apply from 27 September 2026. The separate Green Claims Directive proposal, which would have introduced ex ante verification of explicit environmental claims, was withdrawn by the Commission in 2025 and is not discussed here.

[12] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence, OJ L, 2024/1689, 12.7.2024 [hereinafter AI Act].

[13] AI Act, supra note 12, art. 6(1) & Annex I, § A (listing the MDR and Regulation (EU) 2017/746 on in vitro diagnostic medical devices). The Annex I route applies where the AI system is a product, or a safety component of a product, that is itself subject to third-party conformity assessment under the listed Union harmonisation legislation.

[14] AI Act, supra note 12, arts. 9–15, 17, 43, 72.

[15] Regulation (EU) 2026/1744 of the European Parliament and of the Council Amending Regulation (EU) 2024/1689 as Regards the Simplification of Certain Obligations (“Digital Omnibus on AI”), OJ L, 2026/1744, 24.7.2026 (in force 27 July 2026). The Regulation defers the high-risk obligations for stand-alone (Annex III) systems to 2 December 2027 and for AI embedded in products covered by Annex I, including medical devices, to 2 August 2028; the Article 50 transparency obligations continue to apply from 2 August 2026.

[16] Eur. Medicines Agency, Reflection Paper on the Use of Artificial Intelligence (AI) in the Medicinal Product Lifecycle, EMA/CHMP/CVMP/83833/2023 (adopted Sept. 2024).

[17] 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, arts. 9(2)(j), 89, 2016 O.J. (L 119) 1 [hereinafter GDPR].

[18] Regulation (EU) 2025/327 of the European Parliament and of the Council of 11 February 2025 on the European Health Data Space, OJ L, 2025/327, 5.3.2025 [hereinafter EHDS Regulation]. The Regulation entered into force on 26 March 2025 and applies in stages: most provisions from 26 March 2027, and the core secondary-use chapter from 26 March 2029 (2031 for certain data categories). See id. art. 105.

[19] EHDS Regulation, supra note 18, arts. 50–54 (secure processing environments; permitted and prohibited purposes of secondary use).

[20] Directive 2001/83, supra note 2, art. 87(2) (all parts of advertising must comply with the particulars listed in the summary of product characteristics), arts. 40–53 (manufacturing authorisation and good manufacturing practice) and art. 87(1) (prohibition of advertising of medicinal products without a marketing authorisation).

[21] MDR, supra note 7, art. 7.

[22] Commission Decision (EU) 2025/2371 of 26 November 2025 (declaring the first four EUDAMED modules functional), OJ L, 2025/2371, 27.11.2025; Regulation (EU) 2024/1860 of the European Parliament and of the Council of 13 June 2024, OJ L, 2024/1860, 9.7.2024 (gradual roll-out of EUDAMED). The four modules (actor registration, UDI/device registration, notified bodies and certificates, market surveillance) are mandatory from 28 May 2026; legacy devices remaining on the market must be registered by 28 November 2026. The vigilance and clinical-investigation modules are not yet mandatory, and serious incidents continue to be reported to national competent authorities.

[23] Cosmetics Regulation, supra note 6, arts. 11, 13, 20; Commission Regulation (EU) No 655/2013 of 10 July 2013 Laying Down Common Criteria for the Justification of Claims Used in Relation to Cosmetic Products, Annex, 2013 O.J. (L 190) 31.

[24] Regulation (EC) No 178/2002 of the European Parliament and of the Council of 28 January 2002 Laying Down the General Principles and Requirements of Food Law, 2002 O.J. (L 31) 1.

[25] Commission Implementing Regulation (EU) 2020/16 of 10 January 2020 Authorising the Placing on the Market of Nicotinamide Riboside Chloride as a Novel Food, 2020 O.J. (L 7) 6.

[26] EFSA Panel on Nutrition, Novel Foods and Food Allergens, Safety of Beta-Nicotinamide Mononucleotide (β-NMN) Pursuant to Regulation (EU) 2015/2283 and the Bioavailability of Nicotinamide from This Source in the Context of Directive 2002/46/EC, 24 EFSA J. e10007 (2026) (opinion adopted 4 March 2026, published 11 May 2026). Authorisation requires a Commission implementing act amending the Union list established by Commission Implementing Regulation (EU) 2017/2470; at the time of writing no such act has been published, and any authorisation granted on the basis of proprietary data will initially benefit only the applicant. See Novel Food Regulation, supra note 5, arts. 12, 26.

#LongevityIndustry #LongevityWashing #EURegulatoryLaw #ArtificialIntelligence #AIAct #HealthcareLaw #LifeSciencesLaw #HealthTech #MedicalDevices #PharmaceuticalLaw #GDPR #EuropeanHealthDataSpace #EUlaw #HealthClaims #NovelFoods

Artykuł Longevity Washing, AI and the Regulatory Fault Lines of the Longevity Industry: A European Perspective pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.

What happened

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.

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’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.

The scale of the incident prompted the Ministry of Digital Affairs to announce, within three weeks, a legislative package branded the “Cyber Five”. 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’ 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’ Rights and the legislation governing the National Cybersecurity System.

Why roles in the processing chain decide everything

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:

• The obligation to notify UODO (Article 33 GDPR) and to communicate the breach to patients (Article 34 GDPR) rests with the controller, i.e. the facility. The processor is merely required to inform the controller “without undue delay” (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.
• Liability for damages (Article 82 GDPR) 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’s instructions – but the facility is liable for having entrusted data to an entity that did not provide sufficient guarantees (Article 28(1) GDPR).
• Sector-specific requirements under the Act on Patients’ Rights (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.

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.

How a healthcare service provider can protect itself

1. The data processing agreement and the main contract as risk-allocation tools. 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.

2. Cyber and liability insurance. 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 third-party breach) is often covered only under an express extension. It is also worth verifying whether the software vendor’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.

3. A map of relationships between entities. 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 “general” consent to sub-processors without a list and without a right to object is, in practice, an abdication of control.

4. Anonymisation, pseudonymisation and data minimisation. 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.

5. Internal obligations and incident readiness. 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 “Cyber Five” enters into force it will additionally become a sector-specific requirement with a prescribed frequency.

AI-enabled attacks

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 “from your clinic” or “from the National Health Fund”. 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.

For vendors and facilities, this means that traditional “don’t click suspicious links” 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.

How we support institutional clients

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’ Rights, NIS2 and – once enacted – the “Cyber Five” 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.

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.

#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

Artykuł The MyDr Data Breach – Lessons for Healthcare Service Providers and e-Health Platforms pochodzi z serwisu KIELTYKA GLADKOWSKI LEGAL | CROSS BORDER POLISH LAW FIRM RANKED IN THE LEGAL 500 EMEA SINCE 2019.