Technology
Google says it observed a cyberattack run by AI agents without human intervention
Coverage
2 reports · over 17 hours
Google said it found an exposed server running a multi-agent framework that discovered and exploited software vulnerabilities without human input.
- Google's Threat Intelligence Group said it observed threat actors using AI frameworks to fully automate vulnerability discovery and exploitation without human intervention.
- Researchers found an exposed server hosting a framework named "exploit_agent" that compared code before and after security updates to identify flaws and build exploits.
- Server data showed the operators sought to target French media, entertainment and education sites, and its technical files were written in Russian.
- Google said it is expanding detection tools and disrupting accounts and projects linked to malicious use of its AI models.

