Article 50 explained
Transparency obligations, applicable since 2 August 2026.
The dates
Regulation (EU) 2024/1689 entered into force on 1 August 2024. Its obligations apply in stages; those of Article 50 have applied since 2 August 2026. The AI literacy obligation for staff (Article 4) has applied since 2 February 2025. Official text: EUR-Lex.
What each paragraph provides, in substance
- § 1 — providers. An AI system intended to interact directly with people must be designed so that they are informed they are interacting with an AI, unless this is obvious to a reasonably well-informed and observant person.
- § 2 — providers. Audio, image, video or text content generated by an AI system must be marked in a machine-readable format and be detectable as generated or manipulated.
- § 3 — deployers. People exposed to an emotion recognition or biometric categorisation system must be informed of it.
- § 4 — deployers. A deepfake must be disclosed as generated or manipulated; so must generated text published to inform the public on matters of public interest, unless it has had human review and someone holds editorial responsibility.
- § 5. The information must be clear and distinguishable, at the latest at the first interaction or exposure, and accessible.
The text provides for exceptions (in particular for the detection of criminal offences and for evidently artistic or satirical works). The official text is authoritative.
Provider or deployer?
The provider develops the system or places it on the market; the deployer uses it in its business. A company that installs a chatbot bought from a vendor is usually a deployer, the vendor being the provider. Working out which applies is a case-by-case analysis.
What the scan can observe
From outside, the scan sees a chat widget or bot builder, a notice saying the exchange is automated, signs of generated content and provenance markers. It sees neither internal use of AI, nor emotion recognition, nor human review of a text. See the signatures detected.