In the news · as of 7 November 2026
AI Act Article 50: what a website chatbot has to tell visitors
The EU AI Act’s transparency rules have applied since 2 August 2026. For most company websites they come down to one question: does the chat window on the site tell people they are talking to an AI system, at the start and in plain words?

Photo: Jennifer Bourn, public domain (CC0)
In this guide
Short answer
As of 7 November 2026: Article 50 of the AI Act has applied since 2 August 2026. A chatbot that talks to visitors must tell them, at the latest when the conversation starts, that they are interacting with an AI system. The duty lies with the provider of the chatbot. Fines reach EUR 15 million or 3 % of worldwide turnover, and for small companies the lower of the two.
What does AI Act Article 50 require from a website chatbot?
Article 50(1) of the AI Act (Regulation (EU) 2024/1689) says that AI systems intended to interact directly with people must be designed so that the people concerned are informed that they are interacting with an AI system. A customer-service chat on a website or in a shop is such a system.
Article 50(5) adds how: the information has to be given in a clear and distinguishable manner, at the latest at the time of the first interaction, and it has to meet the applicable accessibility requirements. The European Commission’s guidelines of 20 July 2026 name the moment for a chatbot: when the conversation is launched. A single prominent notice before the first exchange is likely to suffice in most cases.
There is one exception: no notice is needed where it is obvious to a reasonably well-informed person that they are dealing with an AI. The Commission says this exception should be read restrictively, and it names helpdesk chatbots whose replies users may take for human-written as a case where the duty applies. A shop or support chat should not rely on it.
What counts as a clear notice, and what does not
The guidelines give practical examples. A prominent label or banner in plain language, such as “You are interacting with an AI system”, and a first greeting in the chat both work. Some approaches are not enough on their own:
- a generic reference to an “assistant”, without saying that it is an AI system
- a general line such as “Services on this website use AI” on a site that offers several services
- a statement that only names the technology, such as “this system uses LLMs”
- a disclosure that sits only in the terms and conditions
For website chats: avoid a human name and a staff photo for the bot, and put one specific sentence where the conversation starts, for example “You are chatting with an AI assistant. A person from our team can take over.” Keep it visible on phones and readable by screen readers. For chat assistants we build, this notice is part of the first version; see AI assistants.
Provider or deployer: who carries the duty?
The chatbot duty in Article 50(1) is a duty of the provider, the company that develops the system and puts it on the market or into service under its own name. The Commission’s guidelines describe three typical cases, summarised below.
| Your situation | Who is the provider | What you should do |
|---|---|---|
| You built the chatbot in-house and run it under your name | You | Add the notice yourself before the first exchange |
| You embed a vendor’s chat widget without changes | The vendor | Check that the notice appears, ask the vendor to confirm it in writing |
| You changed a vendor’s generative system and run it under your name | You, for the new system | Treat it as your own and add the notice |
Visitors see your website, not the vendor, so check the live page and not only the contract.
What about AI-generated images, video and text?
Two further parts of Article 50 matter for some websites:
- Marking (Article 50(2)). Providers of AI systems that generate synthetic audio, images, video or text must mark the output in a machine-readable way so that it can be detected as artificially generated or manipulated. This concerns companies that offer such a tool. The AI Omnibus, Regulation (EU) 2026/1744, gives providers whose systems were on the market before 2 August 2026 until 2 December 2026 for this duty. It gives no extra time for the chatbot notice.
- Deep fakes and public-interest text (Article 50(4)). Deployers of an AI system that generates or manipulates a deep fake image, audio or video must disclose that the content was artificially generated or manipulated. AI-generated text has to be disclosed only when it is published to inform the public on matters of public interest, and not when it went through human review or editorial control and a person or company holds editorial responsibility. The Commission names a company’s advertising and product descriptions as text outside this rule, although passages with claims about health, consumer safety or sustainability are expressly left out of that example.
How to check your website in five steps
- Find every chatList every chat or assistant window on your website, shop and portal.
- Name the providerNote for each one who built it and under whose name it runs.
- Look at the startOpen each chat on a phone and a desktop and read what a visitor sees before the first answer.
- Fix the wordingReplace vague labels with one plain sentence that says it is an AI system.
- Get it in writingAsk the vendor of an unchanged widget to confirm in writing that its notice meets Article 50.
Fines and who enforces Article 50
Breaches of the Article 50 transparency duties can be fined up to EUR 15 million or 3 % of total worldwide annual turnover, whichever is higher (Article 99(4)). For small and medium-sized companies, and since the Omnibus also for small mid-caps, the lower of the two amounts is the limit. National market surveillance authorities enforce the rules for most systems.
Questions we hear often
Want this sorted without doing it yourself?
Tell us what you are working on. A person replies within one working day.
