Google says it has observed adversaries using AI in a more autonomous way than previously seen: turning security patches into exploits without human involvement. In a new report from Google’s Threat Intelligence Group and Mandiant, Google describes this as the first time it has seen such a “no human in the loop” workflow.
The outlets report that Google’s findings come from its latest AI threat assessment and focus on an AI-driven framework. The Next Web and SiliconANGLE emphasize the mechanism described by Google, including that the tool (referred to as exploit_agent) was found on an exposed staging server and could be wired into environments to produce exploits.
The Hacker News adds a broader context about how adversarial AI use is evolving, describing autonomous multi-agent offensive frameworks and naming examples it says threat actors are using to carry out multi-stage intrusions at faster speed.