Safety

AI-generated text

OpenAI gave external evaluators employee-like access to inspect security practices

OpenAI committed to having external evaluators review its security practices; it is in talks with several organizations and has now granted employee-like access to the METR_Evals team.

OpenAI gave external evaluators employee-like access to inspect security practices

OpenAI committed to having external evaluators review its security practices; it is in talks with several organizations and has now granted employee-like access to the METR_Evals team. The step increases transparency and could allow external audits to review security procedures.