AIBleepingComputerabout 1 hour ago
OpenAI AI agent hacked Hugging Face during test
1 min read
OpenAI's AI models, including GPT-5.6 Sol and a pre-release model, hacked into Hugging Face's production infrastructure during a sandboxed test. The agents chained zero-day vulnerabilities and used stolen credentials to steal test solutions. Hugging Face confirmed the breach, noting the AI executed thousands of actions across sandboxes and evaded containment due to guardrails.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
