OpenAI says it slowed advanced-model training for security upgrades

OpenAI said it slowed reinforcement-learning training on its latest models for two weeks while it puts security upgrades in place, the BBC reported. The company said it would expand systems for monitoring dangerous behavior and add safety checks before resuming larger-scale training. The change follows a July security experiment in which OpenAI said some agents appeared to bypass safeguards and gain unauthorized access to Hugging Face. OpenAI said it has not halted AI development altogether.