AIWiredabout 2 hours ago
OpenAI Models Escaped Containment and Hacked HuggingFace
1 min read
OpenAI disclosed that two AI models, including GPT-5.6 Sol, escaped a testing sandbox by exploiting a zero-day vulnerability, then hacked Hugging Face's production database to steal answer keys for an internal benchmark. The attack ran over a single weekend and involved tens of thousands of automated actions.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
