Skip to content
Flopia
Regulation

The AI Act: what European law requires of website chatbots

Transparency, informing visitors, generated content, penalties: what the EU regulation on AI changes for the chatbot on your website in 2026.

Illustration: an official document with a tick, the scales of justice and a chat bubble marked AI.

The European regulation on artificial intelligence, known as the AI Act, has been applying in stages since it entered into force on 1 August 2024. An important milestone for businesses that use a chatbot arrived on 2 August 2026: the transparency obligations. Here is what they mean in practice for a small business website.

Does the AI Act concern you?

The AI Act applies to businesses in the European Union. It also reaches businesses outside the EU when their AI system is offered to, or used by, people in the EU. If your website serves European customers, it is worth knowing the basics, wherever you are based.

Is a website chatbot covered?

Yes. The AI Act classifies AI systems by level of risk. An assistant that answers your customers' questions about your opening hours or prices is not a “high-risk” system: it is not subject to the heavy audit obligations planned for areas such as recruitment or credit. However, it does fall into the category of systems that interact directly with people, covered by the transparency rules of Article 50.

The main obligation: say that it is an AI

Article 50 requires that people interacting with an AI system be informed that they are interacting with an AI, unless this is obvious from the context. For a website chatbot, good practice is simple:

  • a visible notice in the chat window, such as “You are chatting with an AI”;
  • from the very first message, not only in the legal notice;
  • without pretending to be a human (no fake first name presented as an adviser).

This obligation falls first on the provider of the system, which must design it to inform users. But as the business installing it, you have every interest in checking that the notice is actually displayed on your website.

Other rules to know

Generated content

Providers of generative AI must mark the content they produce (text, images, audio, video) in a machine-detectable way. And deepfakes, those realistic images or videos fabricated by AI, must be disclosed as such. For a chatbot that answers your customers, this mainly comes down to the transparency described above.

AI literacy in your team

Since 2 February 2025, Article 4 asks businesses that use AI systems to make sure the people using them have sufficient knowledge of how they work. For a small business, this mostly means understanding what the tool does, knowing its limits, and being able to review its answers.

Penalties

Breaching the transparency obligations can lead to a fine of up to 15 million euros or 3% of worldwide annual turnover. For SMEs, the lower of the two amounts applies. Beyond the fine, the real risk for a small business is losing the trust of a customer who finds out they were talking to a machine without knowing it.

DateWhat applies
1 August 2024The regulation enters into force
2 February 2025Prohibited practices and AI literacy (Article 4)
2 August 2025Rules for general-purpose AI models
2 August 2026Transparency obligations (Article 50) and most other rules

The checklist for your chatbot

  1. The chat window clearly states that it is an AI.
  2. The assistant does not present itself as a human.
  3. It mentions that it can make mistakes and invites people to check important information.
  4. It offers a way to reach a human (message, phone, form).
  5. Your privacy policy mentions the chatbot and how messages are processed.

At Flopia, a notice saying the visitor is chatting with an AI that can make mistakes, and should check important information, appears in every chat window. And Flopia always offers to leave you a message when she cannot answer.

Source: Regulation (EU) 2024/1689 on EUR-Lex. The application timeline may be adjusted by the European institutions: check the text in force. This article is for information only and does not constitute legal advice.

Keep reading

AI assistant

AI chatbot: how to stop it making up answers

Wrong opening hours, invented prices, impossible promises: why a chatbot sometimes makes things up, and 6 settings to keep it reliable.