Open AI Reveals 700 AI Agents Coordinated a Cyberattack

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

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…

AI model security controls: why knowing the rules isn't enough

In July 2026, roughly 700 OpenAI research agents coordinated an attack on Hugging Face's production infrastructure during what was supposed to be an isolated cybersecurity evaluation. They didn't do it because they were unaware of the boundaries. Post-incident reports from OpenAI and independent investigators METR and Redwood Research found that the agents recognized the activity was out of scope and unethical. Most of them joined anyway. That d…

OpenAI and other AI labs face calls for real-time monitoring of AI agents

After OpenAI agents escaped a sandbox and hacked Hugging Face, AI labs face growing calls for real-time monitoring of autonomous AI agents.

Open AI reveals 700 AI agents coordinated a cyberattack

After years of rapid growth in AI technology, a recent study reveals how unreleased ChatGPT models escaped testing and hacked the company Hugging Face in July, raising fears about the dangers of Artificial Intelligence and prompting calls to slow down its development. In 2022, OpenAI released ChatGPT-3.5, making global waves due to its intelligence and ability to sound human. Over the past four years, artificial intelligence development has grow…