The EU AI Office can now demand access to a model, not just documents about it
From 2 August the Commission's AI Office may request information and documentation, obtain access to models for evaluation, require corrective and risk-mitigation measures, and fine providers up to €15 million or 3% of worldwide turnover. GPAI obligations came into force a year earlier; the enforcement powers were held back for an adjustment period that has now ended.
Access for evaluation is the power that separates this from ordinary disclosure regulation. A regulator that can only read documents is dependent on the accuracy of the documents. A regulator that can obtain the model and test it can check a claim independently — which is a categorically different oversight capability and, so far, a rare one.
The one-year adjustment period is worth noting because it was honoured. Obligations landed in August 2025; enforcement powers activated in August 2026 exactly as scheduled. Regulatory timelines in this field have slipped often enough that a deadline arriving on time is itself information about how seriously this one is being run.
The contrast with the American approach is now unavoidable. One regime publishes its rules and takes the power to inspect. The other convenes voluntary meetings after incidents surface. Both are called AI governance.
European Commission — The enforcement framework of the AI Act → · European Commission — AI Act — regulatory framework → · Beam — EU AI Act 2026: GPAI enforcement and 3% fines begin →