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

发生了什么

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

摘要按规则整理自下方来源原文

来源