Story
As of August 2, 2025, obligations for providers of General-Purpose AI (GPAI) models under the EU AI Act (Regulation (EU) 2024/1689) entered into force. GPAI models placed on the EU market after this date must achieve full compliance by August 2, 2026 — a one-year transition period designed to allow companies to adapt. The European AI Office published the final Code of Practice (CoP) on July 10, 2025, providing voluntary compliance guidance across three pillars: Transparency, Copyright, and Safety and Security. Providers of GPAI models with systemic risk — those exceeding 10^25 floating-point operations per second (FLOPs) during training — face enhanced scrutiny, including conformity assessments by designated Notified Bodies, cooperation with the AI Office, and incident reporting obligations. Non-compliance with GPAI-specific obligations can result in fines of up to €15 million or 3% of global annual turnover. Broader AI Act violations carry penalties up to €35 million or 7% of worldwide annual turnover. The grace period for GPAI models already on the market before August 2, 2025, extends to August 2, 2027. The CoP is not legally binding but offers providers a presumptive pathway to demonstrate compliance with the AI Office, which holds authority to request information and evaluate GPAI models under Articles 91 and 92 of the EU AI Act. Dataconomy reported the core details. Latham & Watkins noted additional context. Data Protection Law Hub added corroborating details.