The study analyses the AI Act's prohibitions on social scoring, predictive policing, emotion recognition, and biometric categorisation, situating each within pre-existing regulation such as data protection and non-discrimination law and CJEU case law concepts including the "essence of fundamental rights" and judicial authorisation requirements. It also addresses the procedural safeguards Member States must establish under the AI Act's "open clauses" — including a legal basis, legitimate aim, and necessity and proportionality safeguards — where they choose to permit exceptions to these prohibitions.
Author: Directorate-General for Communications Networks, Content and Technology (E.J. Kindt)
Status: Adopted / Published
Adoption date: 2026-05-05
Last updated: 19 Aug 2026
Category: Miscellaneous
Subcategory: Report