AI chat must say it is AI
Article 50 of the AI Act applies from 2 August 2026. If your website has an AI chat or publishes AI-made media, this is the part of the Act to read first.
The chat
People who talk to an AI system must be told that it is AI, unless that is obvious from the context. The duty sits with the provider, the company that builds the chat, which has to design it that way. As the company that puts the chat on its site, you want to know that yours does.
Images, audio and video
If you publish deepfakes, which are AI images, audio or video that could pass for real people, objects, places or events, you must disclose that they are artificial. An illustration that misleads no one needs no visible label. The tool that made it must mark its output in a machine-readable way anyway, and that duty is the tool provider’s.
What already applies
Two parts of the Act have applied since 2 February 2025: the ban on prohibited practices in Article 5, and Article 4, which asks companies to make sure the people who use AI for them know enough about it.
What we look at in an audit
- Does the chat say it is AI, in words a screen reader announces?
- Which pictures, sound and video on the site were made with AI, and which of them could be taken for real?
- Which AI tools does the team use, and has anyone been trained?