European Union bans first set of prohibited AI practices

The first substantive prohibitions in the European Union Artificial Intelligence Act began applying on 2 February 2025. Organizations also became subject to a duty to take measures ensuring sufficient AI literacy among staff and others using systems on their behalf.

Prohibited practices include defined manipulative or exploitative systems, certain social scoring, some biometric categorization using sensitive traits, untargeted scraping to build facial-recognition databases and specified emotion recognition in workplaces and schools. Each prohibition contains legal definitions and exceptions; not every biometric or persuasive technology is categorically banned.

Law-enforcement use of real-time remote biometric identification in public spaces is generally prohibited but has narrow authorized exceptions and safeguards. Reporting it as either completely unrestricted or universally impossible would misstate the regulation.

Application did not make all AI Act duties effective at once. Rules for general-purpose models and high-risk systems followed on later schedules, while national authorities and EU institutions developed guidance and enforcement capacity.

The regulation's timetable establishes the milestone. This item distinguishes entry into force in August 2024 from actual applicability of these provisions. A prohibition creates legal exposure and compliance duties; proof of a violation still requires facts about a particular system, purpose and deployment rather than inference from the broad term “AI.”