GuidesAI and compliance

AI Act: what transparency requires from your chatbot

By the Banzai Labs teamUpdated on 2 min read

You are talking to an AITalk to a humanClear noticeFrom the startGenerated content flaggedHuman handoffA chatbot says it is an AI, before the first answer
A chatbot says it is an AI from the start and offers a human handoff.

The obligations, plainly

  • Say it is an AIBefore or at the start of the exchange.
  • Flag generated contentPublished text, images, synthetic voices.
  • Stay understandableA clear sentence, not a hidden clause.

In the interface

Element Example
Welcome message “I am the store’s AI assistant.”
Permanent banner “Answers generated by AI, check important information.”
Visible button “Talk to an advisor”
Link To the privacy policy

Beyond the rule

  1. 1InformThat it is an AI, and about its limits.
  2. 2ProtectThe personal data exchanged (GDPR).
  3. 3Hand offTo a human for sensitive cases.

Transparency is also about trust: a customer who finds out afterwards that they were talking to a machine feels deceived.

The full framework: AI, GDPR and the AI Act. The chatbot: the AI chatbot and test an AI chatbot.

Frequently asked questions

Is a robot icon enough?
A plain sentence, visible before the first exchange, is better: “You are chatting with an AI assistant”. An icon alone may not be understood.
Is my chatbot a high-risk system?
Rarely. A customer service or sales assistant generally falls under the transparency obligations. Uses that decide access to a job, credit or a public service can be classed as high risk.
Does this article replace legal advice?
No. The regulation's application timeline has changed and guidelines keep being published. For a sensitive case, consult a lawyer.

Read next

All guides