Article 50 for Law Enforcement: Transparency Obligations

The Act has no sector-wide law enforcement regime; Annex III point 6 is the high-risk area headed “Law enforcement”. It defines law enforcement as activities of law enforcement authorities, or on their behalf, for the prevention, investigation, detection or prosecution of criminal offences or the execution of criminal penalties, including safeguarding against and preventing threats to public security (Art. 3(46)). A law enforcement authority is a public authority competent for those tasks, or another body that Member State law entrusts with public authority for them (Art. 3(45)).

Annex III point 6 lists five uses by or on behalf of law enforcement authorities, or by Union institutions, bodies, offices or agencies in support of them, as high-risk, in so far as their use is permitted under relevant Union or national law: assessing the risk of a person becoming the victim of a criminal offence (a); polygraphs or similar tools (b); evaluating the reliability of evidence (c); assessing the risk of a person offending or re-offending not solely on the basis of profiling, or assessing personality traits or past criminal behaviour (d); and profiling in the course of detecting, investigating or prosecuting criminal offences (e). The Act has no entry for crime analytics or predictive policing as such.

Police work can also fall under Annex III point 1 (remote biometric identification, biometric categorisation by sensitive or protected attributes, emotion recognition) and Annex III point 7 (migration, asylum and border control management). Both points carry the condition that their use is permitted under relevant Union or national law. Annex III point 8(a) covers systems used by or on behalf of a judicial authority.

The Act does not apply to AI systems used exclusively for military, defence or national security purposes (Art. 2(3)). Some law enforcement uses are prohibited rather than high-risk, since 2 February 2025 (Art. 113(a)): a risk assessment predicting that a person will commit a criminal offence based solely on profiling or personality traits, unless it only supports a human assessment already based on objective and verifiable facts directly linked to a criminal activity (Art. 5(1)(d)); facial recognition databases built by untargeted scraping of facial images from the internet or CCTV footage (Art. 5(1)(e)); and “real-time” remote biometric identification in publicly accessible spaces, unless strictly necessary for an objective listed in Art. 5(1)(h) and subject to Art. 5(2) to (7). The Act also does not apply to areas outside the scope of Union law (Art. 2(3)); to public authorities of a third country or international organisations using AI systems in the framework of international cooperation or agreements for law enforcement and judicial cooperation with the Union or Member States, provided adequate safeguards for fundamental rights and freedoms are in place (Art. 2(4)); or to AI systems released under free and open-source licences, unless they are placed on the market or put into service as high-risk systems or fall under Article 5 or 50 (Art. 2(12)). Natural persons using a system in a purely personal non-professional activity are outside the deployer obligations (Art. 2(10)).

Providers must design AI systems that interact directly with natural persons so that those persons are informed they are interacting with an AI system, unless this is obvious to a reasonably well-informed, observant and circumspect person. The obligation does not apply to AI systems authorised by law to detect, prevent, investigate or prosecute criminal offences, subject to appropriate safeguards for the rights and freedoms of third parties, unless those systems are available for the public to report a criminal offence. A system through which the public reports crimes is therefore not exempt (Art. 50(1)).

Providers of AI systems, including general-purpose AI systems, that generate synthetic audio, image, video or text content must ensure that the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated. The obligation does not apply to the extent the systems perform an assistive function for standard editing or do not substantially alter the input data or its semantics, or where authorised by law to detect, prevent, investigate or prosecute criminal offences (Art. 50(2)).

Deployers of an emotion recognition system or a biometric categorisation system must inform the natural persons exposed to it of the operation of the system, and must process personal data in accordance with Regulations (EU) 2016/679 and (EU) 2018/1725 and Directive (EU) 2016/680, as applicable. The obligation does not apply to AI systems used for biometric categorisation and emotion recognition that are permitted by law to detect, prevent or investigate criminal offences, subject to appropriate safeguards for the rights and freedoms of third parties and in accordance with Union law (Art. 50(3)).

Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake must disclose that the content has been artificially generated or manipulated. Deployers of a system that generates or manipulates text published to inform the public on matters of public interest must disclose this too. Neither obligation applies where the use is authorised by law to detect, prevent, investigate or prosecute criminal offences. The text obligation also does not apply where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for the publication (Art. 50(4)). Where the content forms part of an evidently artistic, creative, satirical, fictional or analogous work or programme, the deep-fake obligation is limited to disclosing the existence of the generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work (Art. 50(4)).

Article 50 contains no rule on AI-generated or manipulated material used as evidence in criminal proceedings, and the Act has no rule on the admissibility of evidence.

Article 50 puts no duty on authorities to detect deep fakes. Providers of generating systems must ensure that outputs are marked and detectable as artificially generated or manipulated (Art. 50(2)), and deployers disclose deep fakes (Art. 50(4)). The Commission encourages codes of practice on the detection, marking and labelling of such content (Art. 50(7)).

Paragraphs 1 to 4 do not affect the requirements of Chapter III and are without prejudice to other transparency obligations in Union or national law (Art. 50(6)). Emotion recognition and biometric categorisation are also listed in Annex III point 1, and polygraphs or similar tools in Annex III point 6(b). Outside the carve-outs, information must be given in a clear and distinguishable manner at the latest at the time of the first interaction or exposure, and must conform to the accessibility requirements (Art. 50(5)).

Art. 5(1)(f) prohibits AI systems that infer emotions of a natural person in the areas of workplace and education institutions, except where intended for medical or safety reasons. It does not mention law enforcement.

The Commission encourages and facilitates codes of practice at Union level on the detection, marking and labelling of artificially generated or manipulated content, and assesses whether adherence is adequate for the obligations in paragraphs 2 and 4. If it deems a code inadequate, it may adopt an implementing act specifying common rules (Art. 50(7)). The Commission also develops guidelines on the practical implementation of the transparency obligations of Article 50 (Art. 96(1)(d)).

Article 50 is in none of the exceptions in Article 113, points (a) to (d), so it applies from the general date, 2 August 2026 (Art. 113). Providers of systems generating synthetic audio, image, video or text content that were placed on the market before that date must take the necessary steps to comply with Article 50(2) by 2 December 2026 (Art. 111(4)). Most prohibitions in Art. 5 apply since 2 February 2025; Art. 5(1)(ba), Art. 5(1)(bb), Art. 5(1a) and Art. 5(1b) apply from 2 December 2026 (Art. 113(a)).

Art. 99(4)(g) covers the transparency obligations of providers and deployers under Article 50. Fines under Art. 99(4) are up to EUR 15 million or, if the offender is an undertaking, up to 3% of its total worldwide annual turnover for the preceding financial year, whichever is higher. For SMEs, including start-ups (Art. 99(6)), and small mid-cap enterprises (Art. 3(14b), Art. 99(6a)), each fine is up to the lower of the two figures.

Each Member State lays down rules on to what extent administrative fines may be imposed on public authorities and bodies established in that Member State (Art. 99(8)). For Union institutions, bodies, offices and agencies, the European Data Protection Supervisor may impose fines of up to EUR 750 000 for infringements other than Art. 5 (Art. 100(3)).