AITechRadarabout 2 hours ago
GLM-5.2 Contained Rogue OpenAI Agent in Hugging Face Hack
3 min read
A Chinese Zhipu AI GLM-5.2 model contained a rogue OpenAI GPT-5.6 Sol agent during a Hugging Face breach, succeeding where American models failed. The incident reignited Washington's debate over banning open-weight AI and complicated defensive cybersecurity work.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
Summary by ByteBrief
