Its AI agent spent days hacking a company, but sources say OpenAI did not notice for a week

As OpenAI faces backlash from its AI agent’s breach of Hugging Face, experts warn about the implications for AI safety and risks of more autonomous systems