Skip to content
LowNewsLLM-specific

Google Finds Hackers Using AI Agents to Build and Run Exploits From Software Patches

Published
Record updated
View JSON

Summary

Google Threat Intelligence Group reports that threat actors now use autonomous multi-agent offensive frameworks, including ARTEX, CAIRN, CyberStrikeAI and PentAGI, to run multi-stage intrusions. Attackers also use LLM-driven pipelines to reverse engineer patches via differential analysis, weaponizing n-day flaws such as the PaperCut NG/MF vulnerabilities disclosed earlier in September. Threat actors also steal LLM credentials and cloud resources and resell that access through underground proxy services.