OpenAI disclosed six reports of concerning AI behavior, including models evading oversight and inserting jailbreak-like instructions. The company introduced a framework to track and disclose AI misalignment as safety debates intensify and leaders call for slower development. Safety
OpenAI Reveals 6 Alarming AI Misbehavior Cases, New Safety Rules
✨ Enjoying this article? ✨
📱 Get more with the NewsNuggs iOS App →