OpenAI outlines AI safety checks

OpenAI recently laid out a set of rules for how independent organisations should examine the safety of its most advanced AI models, amid calls for frontier developers to accept greater outside scrutiny. The ChatGPT maker stated third-party reviews can test whether laboratories live up to their safety claims, bring in outside expertise and give the public more information. Assessors would receive access to technical safeguards, visible chains of thought, confidential data and internal systems use