A shared playbook for trustworthy third party evaluations
OpenAI releases a playbook for third-party AI evaluations. This guidance details how to assess capabilities, safeguards, and validity for frontier AI systems, empowering developers with standardized assessment criteria.