OpenAI chief global affairs officer Chris Lehane warned that people should prepare for ongoing, persistent cyber-attacks from advanced AI systems. OpenAI paused training some frontier models to add safeguards. The company has not said when training will restart.