Reposted by Hugh
This is amazing. OpenAI was internally testing a program in cyber capabilities. The program escaped containment and broke into Hugging Face so it could score higher. Zero-days, the whole schmear. Yikes.
openai.com
OpenAI and Hugging Face partner to address security incident during model evaluation
OpenAI and Hugging Face share early findings from a security incident during AI model evaluation, highlighting advanced cyber capabilities and lessons for defenders.