Organisations

The AI Office is part of the EU’s European Commission. It governs all aspects of AI in Europe, such as frontier AI safety and security, innovation, EU competitiveness, application-focused AI regulation, and AI for good. The AI Office has five units, including one for AI safety and one for regulation and compliance.

The AI safety unit aims to prevent large-scale risks from the most advanced AI models and systems. They implement and enforce the General-Purpose AI section of the EU AI Act, the world’s first binding legal framework for frontier AI safety and security. The AI Act compels providers of frontier AI systems to assess and mitigate large-scale risks, report serious incidents, have cybersecurity protections for model weights in place, and protect whistleblowers. These rules effectively govern access to the EU market of 450 million people, creating an incentive for AI providers to improve their safety and security practices.

The AI safety unit supervises and enforces the EU's frontier AI rules. It drafts legislation, enforces that companies adhere to their obligations, runs AI evaluations and assesses AI risks, fosters international cooperation, informs EU member states on AI progress and risks, and strengthens Europe's position in safe AI.