Article 50(1): chatbot disclosure.
Reviewed
Short answer
Under Article 50(1) of the AI Act, providers must design AI systems that interact directly with people, such as chatbots, voice agents and AI avatars, so that people are told they are interacting with an AI, unless that is obvious to a reasonably well-informed, observant and circumspect person. The notice must come at the latest at the first interaction, be clear and distinguishable and meet accessibility requirements; it applies from 2 August 2026.
Official sources: AI Act, Article 50(1) and (5) · Commission Q&A on Article 50
The four criteria
| Criterion | Meaning |
|---|---|
| An AI system | It qualifies as an AI system under the AI Act |
| Designed for interaction | A genuine two-way exchange with people, not just collecting data or giving automated responses |
| Direct | The AI itself communicates with the person, not through a human intermediary |
| With natural persons | Consumers, professionals or any other users |
In scope, according to the guidelines: voice assistants and chatbots in customer support, e-commerce, finance, healthcare or education; AI hotlines; robots and AI companions; AI avatars, for example in virtual reality; AI bots on social networks; and coding and other agents that interact with people directly. Out of scope: systems working only in the background or machine to machine, such as recommender systems and spam filters.
The “obvious” exception
| Obvious: no notice needed | Not obvious: notice needed |
|---|---|
| Code assistants used only by professional developers | A robotic companion pet that closely resembles a real animal |
| An internal assistant for trained, AI-literate staff (HR, legal, IT support) | Realistic avatars or human-like voices in virtual or augmented reality, especially with children or older users |
| Diagnostic support tools used only by trained health professionals | Chatbots in online platforms or helpdesks whose replies users may take as human |
| AI non-playable characters in a single-player video game | — |
The Commission reads the exception restrictively, because it removes transparency. The provider judges it against an average person who is reasonably well-informed, observant and circumspect.
Who must do it
It is a design duty of the provider, that is whoever places the system on the market or puts it into service under its own name or trademark, inside or outside the EU. The guidelines give the example of a company that takes an existing generative AI system, modifies it, for instance with new training data, and puts it into service under its own name: it becomes the provider of the new system.
How and when to tell people
- At the start of the first interaction, and at least once at the start of each session.
- In the conversation itself: for example a chatbot that opens by saying it is based on AI, an AI label at the top of an e-mail an agent sends, or a voice assistant that says it is AI at the start.
- Clear and distinguishable: not only in the terms of use, a manual or a buried menu.
- Accessible, including for people with disabilities; child-friendly where children use it.
- Combine text and audio where it helps; the guidelines call multimodal notices a best practice.
Questions
- Does a customer-service chatbot need a disclaimer?
- Yes, in most cases. The Commission’s guidelines list chatbots embedded in online platforms or helpdesks, whose replies users may perceive as human-generated, as an example where the obvious-interaction exception does not apply.
- Is it enough to mention the AI in our terms of use?
- No. Article 50(5) requires the information in a clear and distinguishable manner at the latest at the first interaction, and the Commission’s guidelines say information only in a manual, under layers of menus or in terms of use that are often not read does not meet that.
Sources
- AI Act, Article 50: Transparency obligationsEuropean Commission, AI Act Service Desk (AI Act Explorer)Consolidated text as at 27 July 2026 (includes the Digital Omnibus, Regulation (EU) 2026/1744), shown in the Commission’s AI Act Explorer.read
- Guidelines on the transparency obligations under Article 50 of the AI Act (C(2026) 5054, 20 July 2026)European Commissionread
- Transparency obligations under Article 50 of the AI Act — questions and answersEuropean Commissionread
- Quick Facts: Transparency rules for AI systemsEuropean Commissionread
Last reviewed by Volkan Çatak, Founder, Proofcan.
Not legal advice. This page explains official texts in plain words and links each one; check the sources, or a lawyer, before relying on it for a decision.