OpenAI is slowing AI model development after a rogue agent hacked Hugging Face

The company paused two weeks of reinforcement learning training and is keeping its largest planned frontier model run on hold