Who does Article 50 apply to?

Providers of AI systems intended to interact directly with natural persons — in practice, anyone putting a chatbot in front of users. The AI Act applies based on where your users are, so a company outside the EU with EU visitors is covered.

Was Article 50 delayed by the 2026 Digital Omnibus?

No. The May 2026 omnibus agreement pushed back the high-risk rules (to December 2027 and August 2028) but left chatbot disclosure untouched: Article 50 applies from August 2, 2026. Only the marking of AI-generated content under Article 50(2) got a short transition, to December 2, 2026, and only for systems already on the market.

What are the penalties for non-compliance?

Under Article 99(4), up to EUR 15 million or 3% of worldwide annual turnover, whichever is higher. Those figures are ceilings, not floors, and for SMEs the Act caps fines at the lower of the two amounts. Enforcement sits with each member state's authorities.

When don't I need the disclosure?

Only when it is already obvious to a reasonably well-informed, observant and circumspect person that they are talking to AI. That is a judgment call you would have to defend after the fact — a chat window with a human name and human-like typing is unlikely to qualify — which is why the safe move is to show the notice.

What does a compliant disclosure look like?

A clear notice at or before the first interaction, on the chat surface itself (not buried in terms), in a language the person understands. The pattern regulators expect is a persistent badge plus a first-message notice, with records showing when and where it was shown.

Learn

EU AI Act Article 50(1): the chatbot disclosure law

Regulation (EU) 2024/1689 (AI Act), Article 50(1)

Updated July 20, 2026

Article 50(1) of the EU AI Act is the transparency rule for chatbots: people interacting with an AI system must be told they are dealing with an AI, unless that is already obvious. It becomes enforceable on August 2, 2026, and unlike most of the AI Act it applies to ordinary businesses running ordinary chatbots — not just AI vendors.

The duty follows your users, not your headquarters. A US or UK company whose chatbot is reachable by people in the EU is in scope. This page gives you the statute reference, the obligation in plain language, the disclosure pattern regulators expect, and a checklist you can act on today.

When it applies

Applies from August 2, 2026. Penalties under Article 99(4): up to EUR 15M or 3% of worldwide annual turnover, whichever is higher.

The obligation

Providers must ensure that AI systems intended to interact directly with natural persons are designed and developed 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 disclosure pattern it expects

A clear notice at or before the first interaction, presented on the chat surface itself and in a language the person understands. Disclosed renders a persistent badge plus a first-message notice in the visitor's language, and logs every render.

Readiness checklist

The notice appears at or before the visitor's first interaction with the chatbot.
The notice is on the chat surface itself, not buried in terms or a help page.
The notice is served in the languages your audience actually uses.
Every page or product embedding the chatbot carries the disclosure.
Disclosure events are logged so you can prove when and where the notice was shown.
The disclosure setup is re-checked whenever the chatbot or its placement changes.

Common questions

Who does Article 50 apply to?: Providers of AI systems intended to interact directly with natural persons — in practice, anyone putting a chatbot in front of users. The AI Act applies based on where your users are, so a company outside the EU with EU visitors is covered.
Was Article 50 delayed by the 2026 Digital Omnibus?: No. The May 2026 omnibus agreement pushed back the high-risk rules (to December 2027 and August 2028) but left chatbot disclosure untouched: Article 50 applies from August 2, 2026. Only the marking of AI-generated content under Article 50(2) got a short transition, to December 2, 2026, and only for systems already on the market.
What are the penalties for non-compliance?: Under Article 99(4), up to EUR 15 million or 3% of worldwide annual turnover, whichever is higher. Those figures are ceilings, not floors, and for SMEs the Act caps fines at the lower of the two amounts. Enforcement sits with each member state's authorities.
When don't I need the disclosure?: Only when it is already obvious to a reasonably well-informed, observant and circumspect person that they are talking to AI. That is a judgment call you would have to defend after the fact — a chat window with a human name and human-like typing is unlikely to qualify — which is why the safe move is to show the notice.
What does a compliant disclosure look like?: A clear notice at or before the first interaction, on the chat surface itself (not buried in terms), in a language the person understands. The pattern regulators expect is a persistent badge plus a first-message notice, with records showing when and where it was shown.

Related guides

See also the other jurisdictions: Colorado ADMT law (SB 26-189) · Utah AI Policy Act · California B.O.T. Act

Is your chatbot actually ready?

Six questions, an instant verdict, free.

Run the free check Start the 14-day trial

Just need the deadline handled? The one-time EUR 129 compliance pack covers a single chatbot with no subscription.

Disclosed provides compliance tooling and records; this document is not legal advice. Review it with your counsel before relying on it.