OpenAI published the Frontier Governance Framework on May 28, 2026, presenting a public governance document that explains how its safety and security practices align with emerging legal requirements. The company says the framework translates parts of its existing Preparedness Framework into specific obligations under laws such as California’s Transparency in Frontier AI Act and the EU AI Act’s Code of Practice for General Purpose AI.
Scope and purpose
According to OpenAI, the Frontier Governance Framework focuses on regulatory compliance rather than replacing the Preparedness Framework, which the company describes as the foundation for managing the most serious risks posed by advanced AI systems. The published document is framed as a way to map internal practices to identifiable legal and regulatory expectations.
Risks and governance elements covered
The framework addresses risk assessment and mitigation across several specific areas. OpenAI lists cyber offense, CBRN risks, harmful manipulation, and loss of control among the threat categories it considers. It also describes requirements and processes for model reporting, security risk management, incident response, and incorporation of external expert input.
OpenAI states the framework includes provisions for updating governance measures and reporting as capabilities and evaluations evolve. The company also notes that its internal practices include measures that go beyond current legal requirements.
The publication indicates an expectation that the approach will continue to evolve: OpenAI says it will update the Frontier Governance Framework as model capabilities, evaluations, and regulatory requirements develop. The document is presented as available publicly for review and use by stakeholders interested in how the company maps safety practices to specific regulatory obligations.
Original source: OpenAI News