EUAICheck is an educational simulator. We are not a law firm and have no affiliation with any official body.
EUAICheck is an independent educational simulator. This content is informational and does not constitute legal advice.
EU AI Act and chatbots: the transparency obligation has been in force since August 2026
Updated: September 2026 - Article 50 - Regulation (EU) 2024/1689
Since 2 August 2026, if your website uses a chatbot or AI assistant, you must clearly inform users that they are interacting with AI. This obligation is in force. It was not postponed by the Digital Omnibus.
Article 50 of Regulation (EU) 2024/1689 imposes transparency obligations for AI systems that interact with people or generate content. Unlike obligations for high-risk systems (postponed to December 2027), Article 50 has been fully applicable since 2 August 2026.
1. What is Article 50?
Article 50 covers four distinct transparency obligations:
- Disclose that a person is interacting with an AI (chatbots, virtual assistants)
- Mark AI-generated content in machine-readable format (text, image, video, audio)
- Inform people when a system recognises their emotions or categorises them biometrically
- Label deepfakes and AI-manipulated content
2. Who is affected?
According to the European Commission, around 33% of organisations have obligations under Article 50. This notably includes:
- Any website with a customer service chatbot (Intercom, Tidio, Drift, Crisp, Zendesk, etc.)
- Any company using an LLM (ChatGPT, Claude, Gemini) to generate published content
- Any platform publishing AI-generated or AI-modified audio or video content
- HR chatbots for candidate pre-qualification
The obligation applies even if you use a third-party chatbot (Intercom, HubSpot, etc.). As a deployer, you are responsible for displaying the notice. The chatbot provider is not responsible on your behalf.
3. Concrete obligations for your chatbot
- Disclosure must be made at the start of the interaction (first message)
- The notice must be clear and accessible, in the user's language
- The chatbot must not take on a human identity
- The 'obvious from context' exception is very restrictive: a support bot does not qualify
Compliance is straightforward and can be done in a few hours:
- Add a visible notice in the chatbot interface: 'I am an AI assistant' or 'You are interacting with artificial intelligence'
- Configure the chatbot to introduce itself as an AI at the start of the conversation
- Add an 'AI' or 'Powered by AI' badge or label in the interface
- For published AI-generated content: add standardised metadata (C2PA or equivalent)
The European Commission published its final guidelines on Article 50 on 20 July 2026. They specify that the disclosure obligation must be 'prominent, comprehensible and given at the appropriate time'.
5. Penalties for non-compliance
Violations of Article 50 fall under the second penalty tier of the EU AI Act:
- Up to 15 million euros or 3% of annual global turnover
- The national competent authority (CNIL in France) is responsible for enforcement
- Proportional penalties are provided for SMEs
Does your site use an AI chatbot?
Check your situation in 17 questions. Free educational simulator.
Start the simulator