EU Agrees to Postpone Application of AI Regulations

The European Parliament and the Council of the European Union (EU) have reached a provisional agreement to delay the application of regulations for high-risk AI systems. This agreement, part of the 'Digital Omnibus' regulatory package proposed by the European Commission in November last year to strengthen corporate competitiveness, was finalized on the 7th after several trilogue negotiations.

Under this agreement, the mandatory application date for high-risk AI systems, such as those involving biometrics and critical infrastructure, has been postponed by approximately 16 months, from August 2, 2024, to December 2, 2025. For AI systems embedded in regulated products subject to EU safety requirements, the mandatory application date has been extended by approximately 24 months, from August 2, 2025, to August 2, 2028.

Additionally, new provisions have been introduced to prohibit the creation and use of AI systems that depict the physical exposure or sexual acts of identifiable individuals without their consent, and AI systems that generate child sexual abuse material. Companies must bring their relevant systems into compliance with these prohibitions by December 2, 2024.

This provisional agreement is expected to enter into force after formal approval by the European Parliament and the EU Council.