Chatbots: telling visitors they are talking to an AI has been mandatory since 2 August 2026

The EU Artificial Intelligence Act, adopted in 2024, has mostly applied since 2 August 2026. For a website with an AI-powered chat, one rule concerns your visitors directly.

What Article 50 says

Article 50(1) requires providers of AI systems intended to interact directly with people to design them so that those people are informed they are interacting with an AI, unless this is obvious to a reasonably attentive person given the context.

The obligation falls first on the provider of the tool. In practice, as a business installing a chat on your site, you have every reason to choose a tool that clearly presents itself as an automated assistant.

What the Digital Omnibus changed, and what it did not

A website chatbot that answers questions about your offers is not, in principle, a high-risk system. The rule that applies to it is transparency, and it already applies.

Three checks for your chat

• The first message says it is an automated assistant, for example “I’m the firm’s virtual assistant”.
• The displayed name does not suggest a human adviser.
• Handing over to a human is possible and explained: call back, email or booking an appointment.