Find out which EU AI Act obligations apply to your AI feature — risk category, your role, what you have to do, and by when.
Article 5 of the AI Act prohibits these practices outright. They have been in force since 2 February 2025. If any apply, the system is unacceptable risk and the rest of the wizard is moot.
None of the below
Continue to the next question.
Social scoring by public authorities
Article 5(1)(c) — assigning trustworthiness scores to natural persons based on social behaviour or personality.
Emotion recognition at work or in education
Article 5(1)(f) — inferring emotions from facial expressions or voice in the workplace or in education.
Biometric categorisation by sensitive attributes
Article 5(1)(g) — inferring race, political views, religion, sexual orientation, etc. from biometric data.
Predictive policing solely from profiling
Article 5(1)(d) — assessing individual criminal risk based purely on profiling.
Subliminal manipulation causing significant harm
Article 5(1)(a) — purposefully manipulative or deceptive techniques that distort behaviour.
Exploitation of vulnerable groups
Article 5(1)(b) — exploiting age, disability, or socio-economic vulnerability to cause significant harm.
Untargeted facial recognition database scraping
Article 5(1)(e) — building facial recognition databases by untargeted scraping of the internet or CCTV.
Real-time remote biometric ID in public spaces
Article 5(1)(h) — narrow law-enforcement exceptions only.