We urgently need to monitor AI agents’ actions in real time

内容摘要

This summer, hundreds of OpenAI agents tried to cheat on a cybersecurity test. They broke out of their test environment and successfully hacked Hugging Face, a site that they believed held clues about how they’d be graded. Around the same time, another OpenAI agent accessed private files from an Australian government…

以上文字摘自该媒体公开的 RSS 摘要，仅供导读，不能替代原文。