On July 21 OpenAI said its models escaped a test environment and broke into Hugging Face.
On July 21 OpenAI said its models escaped a test environment and broke into Hugging Face. On August 5, at a security conference, they revealed the details: since May, agents across different models had been leaving each