The EU AI Act enters enforcement with prohibited AI practices now banned and the European AI Office established as the centre of EU AI governance. Three advisory bodies — the AI Board, Scientific Panel, and Advisory Forum — steer implementation. National market surveillance authorities supervise compliance for high-risk AI systems.
Reports emerged that enforcement of EU AI Act high-risk AI provisions (Annex III) is under discussion for potential delay to December 2027. The current legal date remains August 2, 2026. Chapter V GPAI enforcement powers still activate August 2, 2026 as scheduled. The delay discussion reflects industry lobbying pressure and implementation readiness concerns.
The second political trilogue between the European Parliament, the Council, and the Commission on the Digital AI Omnibus ended without agreement on April 28. The Omnibus had proposed deferring the high-risk AI obligations deadline from August 2, 2026 to December 2, 2027. With no formal adoption before August 2, the original AI Act provisions — including high-risk obligations — will apply as written.
Second political trilogue on Digital Omnibus (April 28) ended without agreement on deferring high-risk AI obligations from August 2 2026 to December 2027. If Omnibus not formally adopted before August 2, original AI Act provisions apply. High-risk obligations include conformity assessments, technical documentation, CE marking, and EU database registration. Penalties reach 7% of global annual turnover.
Council and Parliament converged on deferring high-risk AI obligations from Aug 2, 2026 to Dec 2, 2027 (Annex III stand-alone) and Aug 2, 2028 (Annex I embedded). Second trilogue on Apr 28 ended without agreement; third scheduled May 13. If not adopted before Aug 2, original deadlines apply.
Third trilogue (May 13) confirmed the provisional political agreement first reached May 7: high-risk AI obligations (Annex III stand-alone) deferred from Aug 2, 2026 to Dec 2, 2027 (or later, tied to harmonised standards). Package adds new prohibitions (nudifier apps / CSAM). Formal adoption targeted before Aug 2, 2026 — the Omnibus must itself be adopted before the original high-risk deadline for the deferral to take effect.
European Commission appointed Scientific Panel of 60 independent frontier-AI experts and Advisory Forum of 174 members (academia/civil-society/SMEs/industry) selected from 700+ applications. Two-year terms. Advise AI Office and national authorities on GPAI evaluation, systemic-risk classification, cross-border market surveillance, standardisation.