AI Act Chatbot Disclosure: Telling Customers They Talk to a Bot
The AI Act chatbot disclosure rule is pretty simple. When people talk to a chatbot, they have to be told there’s a machine on the other end, so they can decide with their eyes open how to carry on. Running a support bot or voicebot on an EU website? This guide covers what the duty includes, when it applies and how to word the notice. You’ll get practical wording examples. Not legal advice, though.
What does the AI Act chatbot disclosure rule require?
The AI Act puts chatbots in the transparency-risk category, which means users must be told they’re dealing with a machine. That category sits between banned practices and minimal-risk tools. According to the European Commission’s AI Act framework page, it covers risks tied to a need for openness about AI use, and the regulation attaches specific disclosure obligations to it. The legal basis? AI Act Article 50. The Commission names two goals here: keeping trust intact and letting people make an informed decision. Fair enough.
When does the chatbot transparency obligation apply?
The AI Act entered into force on 1 August 2024 and became applicable on 2 August 2026. So yes, the chatbot transparency obligation already applies. From that same date the AI Office and Member State authorities implement, supervise and enforce the regulation, as the Commission page linked above explains. The source does mention longer transition periods. But those are for high-risk systems covered by Annex I and Annex III, and a customer service bot isn’t described among them. In my view, leaning on those extensions to put off your notice is a bad bet.
Does the AI Act apply to a small business chatbot?
Yes. The Commission ties the disclosure duty to the use of chatbots, not to how big the company is. A modest bot on a local shop’s site falls under the same idea as one run by a big retailer. (Don’t mix it up with minimal-risk AI, like spam filters or AI-enabled video games, which the regulation leaves alone.) Voicebots work the same way: the caller should know it’s a machine talking. Got an unusual setup? The AI Act Service Desk and the Commission guidelines are the official places to check.
How to disclose an AI chatbot to users: wording examples
What works best is a short, plain notice at the start of the conversation, visible before the user types a single word. It should say what the bot is and, ideally, where its answers come from. A few examples you can adapt:
- “Hi, I’m an AI assistant. I answer questions based on this website’s content.”
- Widget header version: “AI assistant - automated answers”
- “You’re chatting with an automated assistant, not a person. For anything I can’t answer, our contact page lists other options.”
- Voicebot spoken line: “Hello, this is an automated voice assistant. How can I help you today?”
Don’t bury the notice in a footer or the privacy policy. And skip the human names and stock photos that make the bot look like a real person (it fools nobody for long anyway). Keep the tone in line with the rest of the chat, same as your bot replies that sound natural. Multilingual site? Then the message gets translated too, following the principle of one notice, every language.
Checklist before you publish the notice
A quick pass before launch catches most of the gaps in how you disclose an AI chatbot to users. Go through these:
- Place the notice in the first bot message or the widget header.
- Check that it displays fully on mobile screens.
- Translate it for every language version of the site.
- State what the bot can and cannot do.
- Point users to other help, such as a contact page or product FAQ.
- Review the wording again whenever official guidance changes.
If you use Botino, the bot answers from your company’s own content inside a website widget, so the notice can say exactly that. Don’t promise a live person in the chat if nobody will answer there. The notice and where you put it stay your responsibility. Feature details are in our frequently asked product questions.
So AI Act chatbot disclosure really boils down to one clear, early message telling customers they’re talking to a machine. Write your notice today. Put it at the very start of the chat. Then keep an eye on Commission guidance as it develops. Want the bigger picture on automation and regulation? Browse our articles on AI in business.
FAQ
Do I need to tell customers they are talking to a chatbot?
Yes. The Commission describes the AI Act transparency duty as making people aware they’re interacting with a machine when they use a chatbot, so they can make an informed decision. If your setup is unusual, check the official guidance or talk to a lawyer.
Since when does the chatbot disclosure duty apply?
Since 2 August 2026, when the AI Act became applicable. The exceptions the Commission lists concern high-risk systems, not chatbot transparency. So a website bot should already be showing a clear notice.
Is a line in the privacy policy enough?
Probably not. The whole point is that people know they’re dealing with a machine while the conversation is happening. And honestly, who reads a privacy policy before opening a chat? A visible notice in the chat window fits the purpose much better. Check the Commission guidelines for anything more specific on form and placement.
Related posts
What a Chatbot Privacy Notice Should Tell Your Customers
A chatbot privacy notice tells visitors who runs the chat, what they may type into it, why those messages get processed, who…
Chatbot Scams: Helping Customers Tell Your Real Bot From a Fake
Customers can tell your real bot from a fake once you say publicly where the official bot lives and what it will…
What Is a RAG Chatbot and How It Answers From Your Documents
A RAG chatbot is a bot that first searches your own documents for the passages that fit a question, and only then…