Yes. Since 2 August 2026, any chatbot that talks to people in the EU must be designed so that they are told they are interacting with an AI system, at the latest at the first interaction (Article 50(1) and (5)). That duty sits with the chatbot's provider. If the chatbot is used in an Annex III area such as recruitment, credit, insurance or public benefits, it is also a high-risk AI system, with much heavier obligations from 2 December 2027.
What does Article 50(1) actually require?
Regulation (EU) 2024/1689 requires providers of AI systems "intended to interact directly with natural persons" to design them so that those persons are informed they are interacting with an AI system. The information must be clear and distinguishable, given no later than the first interaction, and meet accessibility requirements (Article 50(5)).
The Commission's Q&A on Article 50 sets out four cumulative criteria: it is an AI system, it is built for a genuine two-way exchange, the AI communicates directly rather than through a human, and the other party is a natural person. Background processing and machine-to-machine exchanges are out of scope.
There are two exceptions:
- where it is obvious to a reasonably well-informed, observant and circumspect person, taking the context into account, that they are dealing with an AI. The Commission reads this exception narrowly;
- systems authorised by law to detect, prevent, investigate or prosecute criminal offences, unless they are available for the public to report a crime.
According to the Commission's Q&A, people must be told from the start of the first interaction, in a clear and distinguishable way that meets accessibility requirements. The Commission's Article 50 guidelines, published on 20 July 2026, go into implementation. In practice, an explicit message when the conversation opens, or a spoken statement for a voice assistant, meets this requirement; a line left in the terms and conditions, which people don't read before chatting, hardly does.
My chatbot generates content. Do other duties apply?
Yes, depending on what it produces:
- Article 50(2): providers of systems that generate synthetic text, images, audio or video must mark the output in a machine-readable, detectable format. For systems placed on the market before 2 August 2026, the deadline is 2 December 2026 (Article 111(4), inserted by the Omnibus).
- Article 50(4): deployers who publish deepfakes, or AI-generated text meant to inform the public on matters of public interest, must disclose it. For text, this does not apply where a human has reviewed it and someone holds editorial responsibility.
The Code of Practice on transparency of AI-generated content, published on 10 June 2026, covers paragraphs 2 and 4. It does not cover the chatbot disclosure duty in paragraph 1.
When does a chatbot become high-risk?
High-risk status depends on the use, not the technology. A chatbot is high-risk if it is used in one of the Annex III areas, for example:
- screening job applications or evaluating candidates (point 4);
- assessing creditworthiness or establishing a credit score (point 5(b));
- risk assessment and pricing for life and health insurance (point 5(c));
- deciding admission to education or evaluating learning outcomes (point 3).
Article 6(3) lets a provider rule out high-risk status where the system only performs a narrow procedural or preparatory task without materially influencing the decision, but never where it profiles natural persons. A customer-service bot answering delivery questions is not high-risk; a bot that shortlists candidates is.
Under Regulation (EU) 2026/1744, the "Digital Omnibus", the Chapter III obligations for Annex III systems apply from 2 December 2027 (Article 113). The full set of dates is in our EU AI Act timeline.
Am I the provider or the deployer of my chatbot?
Article 3 defines the provider as whoever develops an AI system, or has it 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. The deployer is whoever uses an AI system under its authority in a professional context.
In practice:
- you wire a third-party model (for example through an API) into a chatbot you launch under your own brand: you are in principle the provider of the chatbot, and the Article 50(1) disclosure is yours to build. The model provider has its own Chapter V obligations;
- you use an off-the-shelf chatbot sold under the vendor's brand: you are a deployer. Disclosure is the vendor's job, but Article 50(4) still applies to certain content you publish, as does Article 4 on AI literacy;
- you are an agency delivering a chatbot to a client: the answer depends on whose name it is put into service under. We cover those scenarios in automation agencies on n8n and Make: provider or deployer?.
The AI Act also reaches providers established outside the EU when the system's output is used in the EU (Article 2(1)(c)).
What should you do now?
- List your chatbots and record, for each one, your role (provider or deployer), the audience and the purpose.
- Show a clear disclosure in the first message, say it out loud for voice agents, and check that it is accessible.
- Check whether the use falls under Annex III; if it does, plan the Chapter III work for 2 December 2027.
- Remove any manipulative technique or anything exploiting users' vulnerabilities, prohibited since 2 February 2025 (Article 5(1)(a) and (b)).
- Keep a dated record of these checks and the measures applied (see how to prove compliance).
Breaching Article 50 can lead to fines of up to EUR 15 million or 3% of worldwide annual turnover, whichever is higher; for SMEs, whichever is lower (Article 99(4) and (6)).
VerifactAI classifies a chatbot from a few declared facts (role, interaction with people, content generation, Annex III area), identifies the applicable obligations and records the controls applied and the evidence.
Sources
- Regulation (EU) 2024/1689 (AI Act), EUR-Lex
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), EUR-Lex
- Article 50, AI Act Service Desk
- Commission Q&A on Article 50
- Commission guidelines on transparency obligations
- Code of Practice on transparency of AI-generated content
This article is general information, not legal advice.