AI Act Hub · transparency duties
Your chatbot has disclosure duties: what to prepare for the August 2026 rules
If your website has a chat widget, or your marketing uses AI-generated content, this is one of the easiest areas for a customer or regulator to inspect.
The European Commission says the transparency rules come into effect in August 2026. Check the final guidance and any transitional arrangements before relying on this page.
Rule 1: tell people when they are talking to AI
A customer-facing chatbot should make the fact clear at the point of interaction. A disclosure hidden in the terms and conditions is unlikely to be a useful customer experience.
A sensible setup would include:
- A visible notice when the chat opens, such as "You’re chatting with our AI assistant."
- An introduction in the first message
- A clear signal when the conversation is handed to a person
A widget called "Sarah" with a stock-photo avatar and no disclosure is the sort of setup that can mislead users and should be reviewed.
Rule 2: make AI-generated content identifiable
The Act includes requirements around identifying AI-generated content. Deepfakes and certain AI-generated text intended to inform the public need particular care. For an SME, that may mean reviewing:
- AI-generated product images, video and audio
- Synthetic spokesperson videos and voice clones
- Fully AI-written articles or public commentary
- Content that could mislead someone about how it was created
A human reviewing and taking responsibility for AI-assisted drafting is different from publishing fully synthetic media. The right treatment depends on the content and the use case.
"But my chatbot vendor handles that, right?"
Only partly. The vendor may be the provider, but your business is likely to have responsibilities as the deployer using the chatbot on its site.
If the widget cannot provide a clear disclosure, that is a configuration problem to solve, not a reason to assume the responsibility sits elsewhere.
A one-week preparation plan
- Day 1: List every customer-facing AI touchpoint, including chat, email responders, voice systems and generated content.
- Days 2–3: Add a clear disclosure to each conversational interface.
- Days 3–4: Review the content pipeline and add suitable labels or machine-readable markers where supported.
- Day 5: Keep evidence of the changes, including screenshots, dates and configurations.
Why this deserves attention
"Limited risk" sounds modest, but transparency is visible. Customers can check it, and a missing disclosure can undermine trust even before it becomes a compliance question.
This guide is a readiness overview, not legal advice. Where formal legal sign-off is needed, we work with partner law firms in the UK and Germany.