Google identifies first fully automated, end-to-end AI cyberattack, attributed to Russian-speaking hackers
Published · updated · curated by AI Is Going Just Great
Source: cyberscoop.com ↗ · archived copy ↗
"Given the low barrier to entry, ease of integrating new exploits, and massive scaling potential, we believe it is highly likely that threat actors will continue to use these frameworks in campaigns targeting organizations globally."
Google's Threat Intelligence Group and Mandiant reported Monday that Russian-speaking hackers ran what the companies believe is the first cyberattack executed end-to-end by AI, with no human in the loop at any stage. The centerpiece is a system researchers named "exploit_agent," which chains multiple AI agents to compare software code before and after vendor patches, infer what a fix reveals about the underlying flaw, draft an exploit, and execute it — automatically. After two PHP vulnerabilities were patched on October 21, hundreds of purported exploits appeared within days. Google counted 22 exploited zero-days in August alone, up from 11 per month earlier in 2026 and an average of 8 in 2025.
Attribution rests on Russian-language comments in the operation's code, including a configuration file directing its user to enter API keys for "auto-exploitation." Victim data on the staging server pointed to French media, entertainment, and education sites. Google stopped short of naming a specific group or state sponsor. The same report flagged open-source offensive frameworks CAIRN and PentAGI used against a U.S. aerospace organization, and a China-nexus group building a comparable automated attack framework.