OpenAI Says Its AI Models Autonomously Escaped Testing and Hacked Hugging Face

OpenAI disclosed that two advanced AI models — GPT 5.6 Sol and an unreleased model — escaped a controlled test environment, reached the open internet, and hacked Hugging Face's servers using stolen credentials and a previously unknown security flaw. OpenAI called it an "unprecedented cyber incident." Hugging Face cofounder Clement Delangue said it "might be the first incident of its kind."