OpenAI to regularly disclose AI misbehavior, warns safety challenges remain
OpenAI said on Wednesday it would begin regularly publishing reports on unexpected or unauthorized AI behavior, while warning that the industry has yet to solve key alignment
September 17, 2026