Unacceptable Risk

The "unacceptable" risk level covers a very limited set of particularly harmful uses of AI that violates fundamental EU rights and are therefore prohibited:

• Exploitation of people's vulnerabilities, manipulation and use of subliminal techniques;
• Scoring or ranking of citizens for public or private purposes;
• Individual predictive policing based solely on personal profiling;
• Random collection of facial images, from the Internet or CCTV footage, to create or expand databases;
• Emotion recognition in workplaces and educational settings, except for medical or safety reasons (e.g. monitoring a pilot's fatigue levels);
• Biometric categorisation of individuals to deduce or infer their race, political opinions, trade union membership, religious or philosophical beliefs, or sexual orientation. Labelling or filtering data sets and categorisation for law enforcement purposes will remain possible;
• Real-time remote biometric identification in publicly accessible spaces by law enforcement authorities, subject to restricted exceptions.