Article 50 AI Act: transparency obligations

This is DryRights editorial content, not legal advice and not an official translation. Every legal statement below cites a specific provision of the official English text of Regulation (EU) 2024/1689 on EUR-Lex, or the full text of Article 50 on DryRights.

1. What started applying on 2 August 2026

Regulation (EU) 2024/1689 entered into force on the twentieth day following its publication in the Official Journal, but applies in stages (Article 113). The general rule is application from 2 August 2026; Chapter IV, which contains Article 50, is not named among the exceptions in Article 113(1)(a)-(c) (Chapters I-II — from 2 February 2025; Chapter III Section 4, Chapter V, Chapter VII, Chapter XII and Article 78 — from 2 August 2025; Article 6(1) — from 2 August 2027), so Article 50 follows the general rule: it applies from 2 August 2026.

2. Provider versus deployer

'Provider' (Article 3, point (3) of the Regulation) means a natural or legal person, public authority, agency or other body that develops an AI system (or a general-purpose AI model) or that has one 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. 'Deployer' (Article 3, point (4) of the Regulation) means a natural or legal person, public authority, agency or other body using an AI system under its authority, except where the AI system is used in the course of a personal non-professional activity. Article 50 assigns different obligations to providers and deployers. Paragraphs 1 and 2 apply to providers. Paragraph 3 and both limbs of paragraph 4 apply to deployers: the first limb concerns deepfakes, while the second concerns AI-generated or manipulated text published to inform the public on matters of public interest.

3. Direct interaction with an AI system

Under Article 50(1), providers must ensure that AI systems intended to interact directly with natural persons are designed and developed so that the persons concerned are informed that they are interacting with an AI system — unless this is obvious from the point of view of a natural person who is reasonably well-informed, observant and circumspect, given the circumstances. The obligation does not apply to AI systems authorised by law to detect, prevent, investigate or prosecute criminal offences, subject to appropriate safeguards for third parties' rights, unless the system is available for the public to report a criminal offence.

4. Machine-readable marking of synthetic output

Under Article 50(2), providers of AI systems, including general-purpose AI systems, generating synthetic audio, image, video or text content must ensure the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated — using technically feasible, effective, interoperable, robust and reliable means, as far as this is technically feasible given the state of the art. The obligation does not apply where the AI system performs an assistive function for standard editing or does not substantially alter the deployer's input data or its semantics, nor within the law-enforcement exception noted above.

5. Emotion recognition and biometric categorisation

Under Article 50(3), deployers of an emotion recognition system or a biometric categorisation system must inform the natural persons exposed to it of its operation, and must process any personal data in compliance with the GDPR, Regulation (EU) 2018/1725 and Directive (EU) 2016/680, as applicable. A separate law-enforcement exception applies, subject to appropriate safeguards for third parties' rights and in accordance with Union law.

6. Deepfakes

'Deep fake' is defined in Article 3, point (60) as AI-generated or manipulated image, audio or video content that resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful. Under Article 50(4), first subparagraph, 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. For content that is part of an evidently artistic, creative, satirical, fictional or analogous work, the disclosure obligation is limited to disclosing the existence of such generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work (see section 9).

7. Text on matters of public interest

The second subparagraph of Article 50(4) sets a separate obligation: deployers of an AI system that generates or manipulates text which is published with the purpose of informing the public on matters of public interest must disclose that the text has been artificially generated or manipulated. This is a distinct requirement from the deepfake rule in the first subparagraph, and it applies to text publications rather than only images, audio, or video.

8. Human review and editorial-responsibility exception

The disclosure obligation for public-interest text (Article 50(4), second subparagraph) does not apply where the AI-generated content has undergone a process of human review or editorial control and where a natural or legal person holds editorial responsibility for the publication of the content. The law-enforcement exception applies here as well.

9. Artistic, satirical and fictional content

For deepfakes (Article 50(4), first subparagraph), a separate, narrower exception applies: where content is part of an evidently artistic, creative, satirical, fictional or analogous work or programme, the transparency obligation is limited to disclosing the existence of such generated or manipulated content in a suitable manner that does not hamper the display or enjoyment of the work — i.e. full disclosure is not required in the same direct way as for a deepfake outside an artistic context.

10. Exceptions and scope limitations

Article 50(5) requires that the information referred to in paragraphs 1-4 be provided to the natural persons concerned in a clear and distinguishable manner at the latest at the time of the first interaction or exposure, and that it comply with accessibility requirements. Article 50(6) clarifies that paragraphs 1-4 do not affect the requirements and obligations set out in Chapter III (high-risk AI systems) and are without prejudice to other transparency obligations laid down in Union or national law. In addition, each of paragraphs 1-4 carries its own law-enforcement or contextual exception as described above — they do not collapse into a single blanket exemption.

11. Enforcement and maximum penalty framework

Non-compliance with the Article 50 transparency obligations falls under Article 99(4)(g) of the Regulation — administrative fines of up to EUR 15,000,000 or, for an undertaking, up to 3% of its total worldwide annual turnover for the preceding financial year, whichever is higher. This is distinct from the highest fine tier under Article 99(3) (up to EUR 35,000,000 or 7% of turnover), which applies exclusively to prohibited practices under Article 5, not to Article 50.

This is a maximum cap, not an automatic fine per violation: Article 99(1) requires penalties to be effective, proportionate and dissuasive, Article 99(6) sets a lower cap for SMEs, and Article 99(7) requires authorities to take into account the nature and gravity of the infringement, the number of affected persons, the degree of cooperation of the infringer, and other factors before setting any specific amount. The penalties chapter (Chapter XII, including Article 99) has applied since 2 August 2025 — meaning it was already in force by the time Article 50 itself started applying on 2 August 2026.

12. Practical checklist

• Identify your role: provider, deployer, or both, for the specific system in question (Article 3, points (3)-(4)).
• If the system interacts directly with people — is it obvious to an ordinary person that it's AI? If not, disclosure is required (Article 50(1)).
• If the system generates audio/image/video/text — is there machine-readable marking of the output (Article 50(2))?
• If emotion recognition or biometric categorisation is used — are the exposed persons informed (Article 50(3))?
• If publishing a deepfake or AI-generated public-interest text — is this disclosed to users, and does the editorial-control or artistic-context exception apply (Article 50(4))?
• Is disclosure provided no later than the first interaction or exposure, in an accessible format (Article 50(5))?

13. Official sources and version date

This page relies exclusively on the official text of Regulation (EU) 2024/1689 from EUR-Lex (CELEX 32024R1689) and on the full text of Article 50 published on DryRights. The articles of the Act have been ingested. Recitals and annexes are not yet fully available in DryRights. The European Commission's guidance on Article 50 and the voluntary Code of Practice on marking AI-generated content are additional sources that clarify but do not replace the Regulation's text, and were not added to DryRights as of this page's publication.

• Official text of Regulation (EU) 2024/1689 on EUR-Lex ↗
• Full text of Article 50 on DryRights
• Browse the Act’s articles in DryRights