EU AI Act Check: Which risk class does your AI system have?
The EU AI Act sorts AI systems into four tiers: prohibited, high-risk, limited risk with transparency obligations, and minimal risk. This check guides you through Art. 5, Art. 6 with Annexes I and III, and Art. 50 in two to three minutes and shows which obligations apply by when. It reflects the law as amended by Regulation (EU) 2026/1744. Your answers never leave your browser.
What risk classes are there?
Prohibited practices under Art. 5, such as social scoring or emotion recognition in the workplace, may not be used at all. High-risk systems under Art. 6 face extensive obligations such as risk management, documentation and human oversight. Limited-risk systems must disclose under Art. 50 that AI is involved. All other systems have no specific requirements beyond the general duty to support AI literacy under Art. 4.
When is an AI system high-risk?
There are two routes. Under Art. 6(1), a system is high-risk if it is a safety component of a product under Annex I that requires third-party conformity assessment, such as a medical device. Under Art. 6(2), it is high-risk if it is used in one of the eight areas of Annex III, for example in recruitment, lending or education.
What does the exemption under Art. 6(3) mean?
A system in an Annex III area is not high-risk if it only performs a narrow procedural task, improves a human result, detects patterns without replacing human assessment, or merely prepares an assessment. The exemption never applies if the system profiles natural persons. Anyone relying on it must document the assessment under Art. 6(4) and still register the system.
Which deadlines apply since Regulation (EU) 2026/1744?
The prohibitions under Art. 5 have applied since 2 February 2025, and the two new prohibitions on intimate deepfakes and abuse material apply from 2 December 2026. The transparency obligations under Art. 50 have applied since 2 August 2026; for generative systems already on the market before then, the marking obligation applies from 2 December 2026. The high-risk obligations were postponed: to 2 December 2027 for Annex III and to 2 August 2028 for Annex I.
Frequently asked questions
Is the EU AI Act Check free?
Yes. The check is free and needs no sign-up. You only need an account if you want to save the result and document it under Art. 6(4).
Are my answers stored?
No. The evaluation runs entirely in your browser. There is no free-text field, no AI evaluation and no transfer to a server. If you copy the link, your answers are contained in the address as a short code.
Does the result replace a legal review?
No. The check gives structured guidance based on your answers. The binding classification is made by your legal function, especially for borderline cases in Annex III.
What is the difference between provider and deployer?
Providers develop an AI system or place it on the market under their own name. Deployers use a system under their own authority. Many organisations are both, for example when they adapt a purchased model and use it internally. In that case choose “Not sure” to see both lists of obligations.
Which legal basis does the check use?
Regulation (EU) 2024/1689 as amended by Regulation (EU) 2026/1744, published in the Official Journal on 24 July 2026 and in force since 27 July 2026. We checked the deadlines against the Official Journal text.