Does the chatbot in my app have to tell people it's an AI?

Article 50(1) of the AI Act says AI systems that talk directly with people must be designed so that people are told they are interacting with an AI system. There is an exception when this is obvious to a reasonably well-informed, observant and careful person, given the circumstances and context (AI Act Service Desk, Article 50; official text on EUR-Lex).

A chatbot built into an app with Lovable, Bolt or a similar tool is the kind of system this covers. The Commission's guidelines give "AI chatbots embedded in online platforms or assistance support tools (helpdesks)" as an example where the obvious-from-context exception does not apply, because users "may perceive" the replies "as human-generated" (Commission guidelines on Article 50, July 2026).

The AI Act uses two words for the businesses involved:

• A provider is a person or organisation that develops an AI system, or has one developed, and offers it or puts it into use under its own name or trademark, whether paid or free (AI Act Service Desk, Article 3).
• A deployer is a person or organisation that uses an AI system under its authority, other than for personal, non-work use (Article 3).

The duty to tell people they are talking to an AI, in Article 50(1), is placed on providers. The Commission's guidelines give an example of a provider that fits many small businesses: a company that "has developed an interactive AI system (e.g. chatbot) in-house and puts it into service in the Union for its own use and under its name or trademark" (Commission guidelines). The guidelines also say providers outside the EU are covered if the output of their AI system is used in the EU.

So if you built a chatbot into your own app and run it under your business name, the guidelines suggest you may be in the provider role for that chatbot. Whether you are depends on your setup, so check with a lawyer if you're unsure.

Deployers have other duties under Article 50, for example telling people about emotion recognition, or labelling deepfakes (realistic fake images, audio or video) (Article 50). Those usually don't apply to a customer-service chatbot.

• 2 August 2026: Article 50, including the chatbot rule, applies. The Commission's timeline says: "Transparency rules (Article 50) start to apply" on this date (AI Act Service Desk timeline).
• 2 December 2026: a later deadline applies only to Article 50(2), the separate rule that AI-generated content must carry a machine-readable mark (a hidden label software can detect). It is for providers whose content-generating AI was already on the market before 2 August 2026 (Article 111(4)).

These dates include the changes made by the "Digital Omnibus on AI", which amended the AI Act. It is now law, not a proposal: Regulation (EU) 2026/1744 of 8 July 2026, published in the EU's Official Journal on 24 July 2026 (Publications Office of the EU). It moved the dates for high-risk AI rules (to 2 December 2027 and 2 August 2028) and added the 2 December 2026 deadline above. It did not move the 2 August 2026 date for the chatbot rule (Article 113).

Article 50(1) has two exceptions (Article 50):

1. It's obvious. You don't have to say it if a reasonably well-informed, observant and careful person would already know they're talking to an AI, given the context. The Commission's guidelines say providers must "assess and demonstrate" this, and they give helpdesk-style chatbots as an example where the exception does not apply.
2. Law enforcement. AI systems authorised by law to detect, prevent, investigate or prosecute crimes, with safeguards. This doesn't apply to ordinary apps.

Article 50(5) says the information must be given "in a clear and distinguishable manner at the latest at the time of the first interaction or exposure", and must meet the applicable accessibility requirements, meaning it must also work for people with disabilities (Article 50).

The Commission's guidelines list things that are not enough on their own (Commission guidelines):

• a mention only in your terms and conditions or other documents
• hidden markers that users can't see
• vague words such as a generic "assistant"
• a general line such as "Services on this website use AI"
• technical wording such as "this system uses LLMs" (large language models, the technology behind most chatbots) that doesn't tell people they are dealing with an AI

The guidelines say a single, prominent notice before the first interaction is "likely to suffice in most instances". They add that riskier settings, such as financial, legal or health advice, complaints handling, or users who may be vulnerable, are likely to need reminders as well.

These are practical suggestions, not legal requirements. They follow the examples in the Commission's guidelines.

1. A first message. Make the chatbot's opening line say what it is, for example: "Hi, I'm an AI assistant. I can help with questions about orders and delivery."
2. A label by the chat box. Show a short line such as "You are chatting with an AI" next to or above the place where people type, and keep it visible during the chat.
3. A clear name. Avoid a human name and photo that could make people think they are talking to a person.
4. Make it accessible. Use real text (not only an image or icon), readable contrast, and wording a screen reader can read out.
5. Keep it after changes. If you ask your AI builder to redesign the chat, check that the label and first message are still there.

Before launch, also check what the chatbot actually says: see how to test the AI chatbot in your app before launch. AnswerTested, our sister product, tests businesses' AI customer-service chatbots for wrong answers and checks AI disclosure.

• Commission guidelines on Article 50. The Commission approved their content on 20 July 2026 (document C(2026) 5054). The approval note says they will be formally adopted once all language versions are available, and only apply from then (Commission library page). We could not confirm that formal adoption has happened yet. Guidelines explain the law. They are not the law itself.
• Code of Practice on Transparency of AI-generated Content. A voluntary code covering marking and labelling of AI-generated content under Article 50(2), (4) and (5). The Commission says it and the AI Board "have confirmed that the code is an adequate voluntary tool to demonstrate compliance", and that about 190 organisations had signed it by the end of July 2026 (Commission code of practice page). It covers AI-generated content, not the chatbot rule in Article 50(1).