EU AI Act Article 50, explained clause by clause

Article 50 of Regulation (EU) 2024/1689 (the EU AI Act) sets out transparency obligations for providers and deployers of certain AI systems. Under Article 113, it applies from 2 August 2026. Unlike the high-risk rules, Article 50 catches very ordinary products: customer-support chatbots, AI writing and image tools, and any published synthetic media. All quotes below are from the Official Journal text on EUR-Lex.

Article 50(1) — Tell people they are talking to an AI

“Providers shall ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural 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…”

If you run a chatbot, voice bot or any conversational AI, users must be informed it is an AI. The only exemptions are where this is genuinely obvious in context, and certain law-enforcement uses. The safest interpretation for a commercial product: show an explicit disclosure at the start of the interaction. For a product-by-product walkthrough of this test, see is my chatbot covered?

Article 50(2) — Mark synthetic content machine-readably

“Providers of AI systems, including general-purpose AI systems, generating synthetic audio, image, video or text content, shall ensure that the outputs of the AI system are marked in a machine-readable format and detectable as artificially generated or manipulated.”

This is a provider duty. Technical solutions must be “effective, interoperable, robust and reliable as far as this is technically feasible”. In practice this points to provenance metadata standards such as C2PA Content Credentials and watermarking. It does not apply where the AI performs an assistive function for standard editing or does not substantially alter the input data or its semantics.

Article 50(3) — Emotion recognition & biometric categorisation

Deployers of emotion recognition or biometric categorisation systems must inform the exposed persons of the system's operation, and process personal data in accordance with the GDPR (Regulation (EU) 2016/679) and related EU data protection law. Limited law-enforcement exemptions apply.

Article 50(4) — Deep fakes and AI-written public-interest text

Two separate duties for deployers:

• Deep fakes: anyone deploying an AI system that generates or manipulates image, audio or video content constituting a deep fake “shall disclose that the content has been artificially generated or manipulated”. For evidently artistic, creative, satirical or fictional works, disclosure may be done in a way that does not hamper the display or enjoyment of the work.
• Public-interest text: AI-generated or manipulated text “published with the purpose of informing the public on matters of public interest” must be disclosed as AI-generated — unless it has undergone human review or editorial control and a natural or legal person holds editorial responsibility.

Article 50(5) — How and when to disclose

“The information referred to in paragraphs 1 to 4 shall 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. The information shall conform to the applicable accessibility requirements.”

Three tests for any disclosure: clear and distinguishable, delivered no later than first interaction/exposure, and accessible. A footer note buried below the fold is unlikely to qualify.

Article 50(6) and 50(7) — Interplay and codes of practice

Paragraph 6 clarifies these duties sit alongside (not instead of) the high-risk requirements of Chapter III and other EU/national transparency law. Paragraph 7 tasks the AI Office with facilitating codes of practice for detection and labelling of artificially generated content, which the Commission may approve — or replace with common rules by implementing act.

What should you do now?

1. Map which of your features interact with people or generate content.
2. Determine your role (provider, deployer, or both) per feature.
3. Add visible disclosures and machine-readable marking where required.
4. Document your reasoning — including any exemption you rely on.