Assessment.Walk every branch of the regulation.

This covers AI that influences people without their awareness or through intentionally misleading methods — hidden persuasion, dark patterns, or deceptive framing that undermines free choice.

This applies if the AI targets people who may be less able to resist — children, elderly, people with cognitive disabilities, or those in financial distress — to change their behaviour in harmful ways.

This covers AI systems that build profiles of people based on their social conduct or personality and then use those profiles to deny them services or opportunities unrelated to the original context.

This prohibits AI that predicts whether someone will commit a crime based only on who they are — their appearance, demographics, or personality — rather than specific observable behaviour.

This prohibits building facial recognition training data by mass-collecting images from the web or surveillance cameras without the knowledge or consent of the people depicted.

This covers AI that tries to read how employees or students feel — through facial analysis, voice tone, body language — in work or school contexts, unless it serves a medical or safety purpose.

This prohibits live facial recognition or other biometric identification in public places by police, except for a small set of serious crimes and missing persons situations defined in the Act.

This prohibits AI that analyses physical or behavioural traits to deduce sensitive characteristics like ethnicity, political views, religion, or sexual orientation.