Prohibited AI practices are ways of using artificial intelligence that European regulation considers incompatible with fundamental rights and does not permit even with the consent of the data subjects. These include, in particular, the use of subliminal or manipulative techniques causing significant harm, exploiting the vulnerability of persons due to age, disability, or social situation, social scoring of people leading to unjustified disadvantage, and certain forms of biometric categorisation and emotion recognition, for example in the workplace. For an ordinary company, the last is the most relevant: deploying tools that assess employees' emotions, or scoring clients across unrelated areas. If you are considering such a use, stop before the pilot and have it reviewed legally, since the penalties in this category are the highest.
See also: AI risk classification, Internal AI usage policy, Fundamental rights impact assessment.
