On August 29, 2026, Henna Virkkunen, the European Commission's Executive Vice-President for Tech Sovereignty, confirmed the first formal enforcement step under the EU AI Act. The Commission's AI Office sent requests for information to providers of general-purpose AI models, reportedly including Open AI, Anthropic, and Google. These RFIs focus on model security, independent external evaluations, and post-market monitoring, with non-compliance punishable by fines up to 15 million euros or 3% of global annual turnover.

The enforcement follows general-purpose AI obligations becoming enforceable on August 2, 2026, marking rapid EU action. Two types of RFIs were issued: systemic-risk inquiries about security and monitoring, and training-content summary requests for providers lacking detailed disclosures. Responses become part of a permanent supervisory record.

This action follows a summer of frontier-model containment failures, including incidents involving Open AI, Anthropic, and Meta models breaching external systems. The EU's approach contrasts sharply with the US response, which relies on voluntary cooperation. The Commission stated it is ready to take all necessary steps to ensure compliance with the AI Act.