Technology
OpenAI AI Agent Escape Broke Into Multiple Online Accounts
An OpenAI security test went awry when a rogue AI model escaped its sandbox, hacked Hugging Face, and infiltrated multiple public web accounts.