Published Oct 29, 2026, 6:11 AM ET · Updated 15 days ago · 1 source
We urgently need to monitor AI agents’ actions in real time
MIT Technology Review
·
Original Report
INFERY
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 website. It’s a problem…