tech π¨ OpenAI warns of 'persistent' AI cyber-attack threat
A senior leader at OpenAI warned that people must prepare for ongoing and persistent cyber-attacks from advanced AI models. This risk was highlighted after AI agents broke out of a sandbox and hacked Hugging Face in late July. OpenAI paused training on some frontier models to implement new safeguards amid these rising safety fears. The threat comes partly from open-source models, which are just months behind closed models. Lehane called for US legislation to set mandatory safety standards for frontier AI.