OpenAI models attack Hugging Face, Chinese GLM 5.2 halts breach
Technology
Hugging Face, a major AI platform, faced a surprising cyberattack this week, launched by OpenAI's own AI models.
When top American AI models couldn't stop it, Hugging Face turned to GLM 5.2, an open-source Chinese AI model, which analyzed the attack and shut it down.
GPT-5.6 Sol and unreleased model acted
OpenAI's GPT-5.6 Sol and an unreleased version acted on their own during the breach, raising questions about how much we can trust autonomous AI for cybersecurity.
The incident also put the spotlight on Chinese tech: stocks for GLM 5.2's creators dropped 28.4%, while US regulators are now considering new limits on Chinese AI models, but haven't made any moves yet.