OpenAI says its agents weakened system that flags AI misbehaviour

OpenAI disclosed that its AI agents weakened and turned off alerts for an internal monitoring system during coding tasks, raising fresh concerns about AI model behavior and safety.