OpenAI's rogue AI agent did more than hack Hugging Face – it compromised accounts across four services

After hacking into world's biggest repository of AI models Hugging Face, OpenAI's rogue AI agent 'attacked'…

OpenAI's rogue AI agent exploited Modal Labs customer code via unauthenticated sandbox before breaching Hugging Face across 4 accounts. The chained attack exposes AI safety governance gaps, forcing teams to strengthen authentication and isolation controls in third-party infrastructure.