Supporting the Development of Joint Standards for Advanced AI

Advanced AI models can accelerate cyber defense and scientific discovery, but without adequate safeguards, they can also pose serious security risks. To address this, society needs institutions with the technical and governance capacity to evaluate and manage these models.

OpenAI is supporting the establishment of the Appia Foundation, hosted by the Linux Foundation. Appia develops open, modular specifications that translate international standards and frameworks into practical evaluation criteria. The goal is to enable compatibility checks between models, infrastructure, and applications developed by different organizations, and to build a trusted technical language.

OpenAI contributes to this standardization effort by operating the following systematic safety infrastructure:

• Preparedness Framework: An internal operating approach that defines and manages the most severe risks of advanced AI systems
• Frontier Governance Framework: A public governance document focused on regulatory obligations such as risk assessment, model reporting, security controls, and incident response
• Third-party evaluation partnerships: Testing partnerships with US CAISI and UK AISI have achieved practical system improvements, including preventing biological misuse

OpenAI also contributes to standardization and pre-standardization work across the AI ecosystem by participating in various international standards bodies, including ISO/IEC JTC 1/SC 42, the NIST AI Consortium, and the Frontier Model Forum.