What is the EU AI Act? Definition for website owners

The EU AI Act is the European Union's law on artificial intelligence. It sorts AI systems by risk and sets duties for the companies that build and use them.

How it works

The Act, Regulation (EU) 2024/1689 (EUR-Lex), uses four levels of risk, as the European Commission summarizes them (European Commission):

• Unacceptable risk: banned practices, such as manipulative or exploitative systems.
• High risk: uses that can seriously affect health, safety or rights, such as CV sorting or credit scoring. These carry strict obligations before going on the market.
• Transparency risk: systems where people must know AI is involved, for example that "humans should be made aware that they are interacting with a machine".
• Minimal risk: most other AI, such as spam filters, with no new rules.

It also separates providers, who build an AI system or put it on the market, from deployers, who use one in their business.

The timeline, as of October 2026:

The high-risk dates moved under the "AI Omnibus", Regulation (EU) 2026/1744, which entered into force on July 27, 2026. It left the August 2, 2026 start of Article 50 in place, with a grace period until December 2, 2026 only for the duty to mark AI-generated content in systems already on the market (Goodwin).

Why it matters for your website

A chat assistant on a website is an AI system that interacts directly with people. Article 50(1) requires such systems to be designed so people are told they are dealing with an AI, unless that is obvious, and Article 50(5) says this must happen "in a clear and distinguishable manner at the latest at the time of the first interaction" (AI Act Service Desk). Goodwin notes that transparency breaches can be fined up to 15 million euros or 3% of worldwide annual turnover.

The duty is formally on the provider, usually your vendor, but you control the greeting and the widget, so make the disclosure visible yourself. Most website assistants are not high-risk, but some uses can be: screening job applicants, for example, falls under employment. The Act sits alongside the GDPR, which still governs any personal data the assistant collects.

This is a summary, not legal advice. Our article on AI chat and privacy covers what to tell visitors.

Example

A B2B site's assistant opens with: "Hi, I'm an AI assistant. I answer from this website's pages and can pass your question to the team." The disclosure is visible before the first reply, readable by screen readers, and repeated in the widget's title.

Related terms

• GDPR: the EU's data protection law, which applies alongside the AI Act.
• AI agent: the kind of system many website assistants now are.
• Guardrails: controls that keep an AI system within its rules.
• Human handoff: passing a conversation to a person.
• Large language model (LLM): the general-purpose models the Act also regulates.