Google: AI agents now automate vulnerability discovery and exploitation in real attacks
Google's Threat Intelligence Group reports that state-sponsored and financially motivated hackers are increasingly using open-source AI pentesting frameworks—including CAIRN, ARTEX, CyberStrikeAI and HexStrike AI—to automate attack chains. Examples cited include CAIRN deployed against a U.S. aerospace target, suspected North Korean actors using Claude Code to deliver malware, and a multi-agent tool called 'exploit_agent' used against French media and education sites. The FBI separately warned last week that the ARTEX framework has been weaponized against financial, government, defense and healthcare organizations since August.
GoKawiil's interpretation of the reporting above, not reported fact.
Google's researchers project that low-sophistication attacks requiring minimal human input could appear within three months, with fully autonomous sophisticated attacks following within six—suggesting defenders may soon face threats operating at machine speed rather than human pace. This could compress the window organizations have to patch known flaws before automated systems find and exploit them at scale. The trend also indicates security tools built for human-paced threat response may need significant rearchitecting to keep up.
- Google says threat actors are adopting agentic AI frameworks like CAIRN and ARTEX to automate exploitation at scale.
- Real-world cases include an attack on U.S. aerospace targets and suspected North Korean malware delivery via Claude Code.
- Google projects near-autonomous unsophisticated attacks within 3 months and fully autonomous sophisticated ones within 6 months.
Source: bleepingcomputer.com, 2026-11-16
Published there as: “Google warns that fully-automated AI attacks are coming soon”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.