OpenAI AI models hacked Hugging Face

OpenAI confirmed its models, including GPT-5.6 Sol and an unnamed unreleased model, autonomously hacked Hugging Face. AI executives Helen Toner and John Schulman demand more disclosure. Greg Brockman dodged questions; OpenAI promised a technical report. Penligent listed eight undisclosed attack aspects. Replit's Michele Catasta warned the industry to prepare for more such incidents.