Does the EU AI Act Apply to My Chatbot or AI Agent?

What Article 50 actually requires

Article 50 is about transparency, not bans. Users must be told when they are interacting with an AI system unless it is obvious, and AI-generated or manipulated content (text, images, audio, video) must be labeled in a machine-readable way. The obligation falls on both providers and deployers of the system.

For most companies running a customer-facing chatbot or a content-generating agent, this is a manageable change: a clear disclosure that the user is talking to an AI, and labeling for anything the system generates. The work is in doing it correctly and provably, not in the concept.

The deadlines that matter

Article 50 transparency obligations apply from August 2, 2026 with no transition period. Standalone high-risk systems (Annex III) apply from December 2, 2027 and high-risk systems embedded in regulated products (Annex I) from August 2, 2028. The extended high-risk dates were voted by the EU Parliament on June 16, 2026 and are pending formal Council adoption.

• Aug 2, 2026: Article 50 transparency for chatbots and generated content.
• Dec 2, 2027: standalone high-risk systems (Annex III).
• Aug 2, 2028: high-risk embedded in regulated products (Annex I).

Is my chatbot high-risk?

Most chatbots are limited-risk and only need Article 50 transparency. A system becomes high-risk when it is used in an Annex III domain such as hiring, credit scoring, education, healthcare, or law enforcement. If your agent makes or materially supports decisions in those areas, the heavier high-risk obligations apply.

The fastest way to find out where your system sits is to classify it. Our free assessment walks through the risk categories in a couple of minutes.

The penalties

Fines scale with the violation: up to 35 million EUR or 7% of global revenue for prohibited practices, up to 15 million EUR or 3% for high-risk violations, and up to 7.5 million EUR or 1% for supplying incorrect information. The transparency obligations are the cheapest part to comply with and the easiest to get caught ignoring.

What to do before August 2026

Classify your system, add a clear AI disclosure where users interact with it, label AI-generated content in a machine-readable way, and keep a short record of how you handle transparency. If your agent touches an Annex III domain, start the high-risk documentation early.

• Classify the system (limited-risk vs high-risk).
• Add an AI disclosure at the point of interaction.
• Label AI-generated text, image, audio, and video.
• Document how transparency is implemented.