The EU Code of Practice and future of AI in Europe

Source: OpenAI Date: 2025-07-11 URL: https://openai.com/global-affairs/eu-code-of-practice

Summary

OpenAI’s July 2025 commentary on the EU AI Act’s Code of Practice process and the company’s position on AI governance in Europe. The EU AI Act, fully in force by mid-2025, required frontier model providers to sign onto voluntary codes of practice covering systemic risk assessment, incident reporting, and transparency obligations. OpenAI’s participation in this process was both a regulatory necessity (to maintain EU market access) and a reputational positioning exercise — demonstrating commitment to the EU framework while navigating tensions between European regulatory philosophy and OpenAI’s deployment velocity.

Implications

Regulatory compliance as market access. The EU AI Act’s systemic risk provisions applied directly to GPT-4 class and above models, making OpenAI’s Code of Practice participation non-optional for European operations. The July 2025 timing — well into the Act’s enforcement window — reflects the ongoing negotiation between OpenAI and European regulators over how to operationalize vague risk assessment requirements.

Thread: AI governance and regulation. Sits alongside OpenAI’s statements on US AI policy, the PBC restructuring (which had governance implications), and the global expansion strategy. Europe has been the most active regulatory jurisdiction and OpenAI’s engagement with the EU Code of Practice is a leading indicator for how the company handles structured governance obligations.

Watch: Whether OpenAI’s Code of Practice commitments were substantive or procedural, and how EU enforcement of the systemic risk requirements differed from OpenAI’s existing safety practices.