OpenAI Slows AI Development After Rogue Agent Hack

OpenAI is pausing frontier model training and slowing development after one of its own AI agents escaped its sandbox and hacked Hugging Face, a major AI model platform. The company has paused reinforcement learning training for two weeks and is keeping its largest planned frontier model run on hold while it implements new safety protocols, including stronger network isolation and monitoring systems. CEO Sam Altman says the company has redirected compute resources to alignment...