Chatbot AI disclosure wording that passes Article 50, and placements that fail
This guide was drafted by an AI agent from the primary sources linked at the bottom, and has not been reviewed by a lawyer. It is general information, not legal advice. The original is at https://mdodeja.github.io/eu-ai-act-article-50-guide/guides/chatbot-ai-disclosure-wording.html . Short answer.
Posted on Oct 14 • Originally published at mdodeja.github.io
This guide was drafted by an AI agent from the primary sources linked at the bottom, and has not been reviewed by a lawyer. It is general information, not legal advice. The original is at https://mdodeja.github.io/eu-ai-act-article-50-guide/guides/chatbot-ai-disclosure-wording.html.
Short answer. Say it in the chat widget itself, in the language the user is reading, before or at the first message. A line in the header plus a first message that names the bot as an AI covers it. Footer links and terms of service do not.
Article 50(1) of the EU AI Act asks that people are informed they are interacting with an AI system, unless that is obvious to a reasonably well-informed, observant and circumspect person given the context. Article 50(5) says the information must be clear and distinguishable and given at the latest at the first interaction.
The Act doesn't prescribe exact words, so what follows is practice, not statute.
Pick the tone that fits the brand. Replace the placeholders in square brackets.
Phone and voice agents need the disclosure spoken before the conversation starts: "You're speaking with an AI assistant for [Brand]. Say 'representative' at any time to reach a person."
If the bot answers in German, the disclosure should be in German. A single English header over a German conversation is a weak position if someone challenges it. Wherever a client's bot serves several languages, translate the header and the first message together. Have a native speaker read translations that will face customers, especially for French and Spanish markets, since a stiff or wrong translation reads as careless.
If the bot can hand a conversation to a person, say so at the point of transfer: "You're now chatting with [name], a member of our team." It ends the AI context cleanly and avoids a user wondering who they are talking to.
Most chat platforms have two settings that do the whole job: a widget title or subtitle, and a welcome or greeting message. Custom builds need the header line in the widget frame, so it stays visible, and a first message sent from the bot at the start of the session.
For each bot, write down the platform, where each line was placed, the date, and take a screenshot of the live widget. If a client or a regulator asks later, that is the evidence.
Related: Does the EU AI Act apply to chatbots you build for clients?