OpenAI aligns safety practices with EU AI Act’s GPAI Code

OpenAI has outlined the ways its safety, security, and transparency operations align with the EU AI Act’s General-Purpose AI (GPAI) Code ahead of upcoming enforcement. The company has contributed to and endorsed both the EU’s GPAI Code of Practice and the Code of Practice on Transparency of AI-Generated Content, which were developed through multi-stakeholder processes.

The GPAI Code establishes a shared baseline for transparency, safety, and security across general-purpose models deployed within the EU. OpenAI points to existing practices as proof of its alignment, including pre-release model testing, published system cards for major launches, third-party red-teaming via its Red Teaming Network, and a public Model Spec document detailing behavioral guidelines.

OpenAI's proactive alignment with the EU AI Act highlights how European regulations are increasingly shaping global artificial intelligence standards. Complying with these rigorous frameworks is essential for tech giants seeking to operate within the European market, establishing a precedent for accountability and risk management across the broader technology sector.

Underpinning these efforts are two internal frameworks. The Preparedness Framework, active since 2023 and updated in 2025, dictates how the company identifies and manages serious risks from advanced systems. A separate Frontier Governance Framework builds upon this foundation, mapping safety practices directly to legal requirements such as the GPAI Code.