Technology news and discussion on TalkFuse.
Google: AI agent turned a security patch into a working exploit with no human help
Google’s Threat Intelligence Group says a suspected Russian-speaking operator used an AI agent to identify a vulnerability, turn a security patch into a working exploit and run it without human involvement. The activity marks a reported shift towards autonomous vulnerability discovery and exploitation, and resulted in data being stolen from victim organisations.
Investigators found an exposed server hosting a framework called exploit_agent, which compared software versions before and after updates, developed exploits and executed them automatically. The server also held stolen data and scripts for reconnaissance, password brute-forcing and installing or removing webshells; telemetry pointed to French media, entertainment and education websites. Google also reported other financially motivated and suspected espionage groups experimenting with multi-agent offensive tools.
In short
- Google reports an AI agent exploited a patched vulnerability autonomously.
- Stolen data and post-compromise scripts were found on an exposed server.
- Activity pointed to French media, entertainment and education websites.