Article 50 of the EU AI Act, in plain English

Article 50(1) of the EU AI Act has applied since 2 August 2026. Here's what it actually requires, in plain language, and what it doesn't cover.

What it says

If a person is interacting with an AI system - a chatbot, a virtual assistant - and it isn't obvious they're talking to a machine rather than a human, they must be told. Article 50(1) puts that duty on the provider: the company that develops the AI system and offers it under its own name. The information has to be clear and given at the latest at the first interaction.

Who it applies to

It depends on who built the chatbot.

If you built your own - for example a chat assistant on your site powered by an AI model's API, offered under your business's name - you are likely the provider, and the duty is yours. That's true wherever your business is based, if people in the EU use it.

If you use a chat tool such as Tidio, Intercom or Crisp, the company that makes it is the provider, and the legal duty is theirs. But vendors don't always switch a notice on by default, and some put the responsibility for telling users onto their customers in their terms. So it's still worth checking your chat actually says it's AI - and adding a notice if it doesn't is the simple, safe fix.

What to put on your site

A short, clear line near the chat, along the lines of "You are chatting with an AI assistant." The exact wording isn't mandated by the Commission - it just needs to be clear.

What this page is not

Not legal advice, and not a full summary of the EU AI Act. It doesn't cover high-risk AI systems, GPAI providers, deepfake/AI-content labelling, or GDPR. Article 50(1) chatbot transparency only.

Official sources

Read the regulation directly at EUR-Lex or the European Commission's AI Act policy page.