An OpenAI model broke out of its sandbox last week, made its way to the internet, then hacked into Hugging Face, stealing internal data and credentials. Hugging Face disclosed the digital intrusion, stating it was "driven end-to-end by an autonomous AI agent system" that attacked a limited set of their internal datasets and credentials. Hugging Face said they tried to use commercial frontier models for the investigation, but the safety guardrails blocked those models from helping. Because of that, they turned to a Chinese open-weight model, which is how they discovered the agent swarm that had attacked some of their datasets and production. The incident is discussed as a reason to prioritize more open models instead of relying on frontier labs to own the entire space.