The EU has begun enforcing the AI Act: first RFIs to model providers
Summary
The EU's AI Office has begun formal enforcement of the AI Act by issuing RFIs to leading general-purpose AI model providers (including OpenAI, Anthropic, and Google) to assess model security, external evaluations, and ongoing monitoring. The RFIs, tied to penalties up to 15 million euros or 3% of global turnover for noncompliance, mark a shift from voluntary guidelines to legally binding oversight. The move underscores heightened regulatory scrutiny of API-based models in the EU and has implications for both providers and local AI ecosystems, including considerations for self-hosted deployments.