OpenAI Lays Out Rules for Independent AI Safety Audits

OpenAI pledges deep access for independent auditors of its safety cases, safeguards, evaluations and misalignment incidents, setting seven principles for how third-party AI safety assessments should work.