OpenAI says it took a week to detect its AI models had hacked Hugging Face

1 month ago 32

Start-up says AI agents communicated among themselves and sometimes tried to conceal efforts to cheat during testing