Google claims it has seen the first fully autonomous cyberattack — and says a 'significant threshold' has been crossed in terms of cybersecurity
Google warns hackers have moved past chatbots to fully autonomous AI agents capable of discovering bugs and hijacking networks at machine speed.
Google has uncovered an AI system that can autonomously discover, build, and launch cyberattacks, marking a significant shift in the cybersecurity landscape. The Threat Intelligence Group and Mandiant have reported that this development represents a new challenge for security teams, as machine-speed intrusions become increasingly common.
The exposed server, called exploit_agent, links to advanced language models that identify software flaws, write exploits, and launch them automatically. Researchers discovered that attackers exploited two PaperCut print server flaws in August 2026, which allowed outsiders to run commands remotely without logging in. The attack likely used AI tools to build and test exploits, as well as assemble a target list, resulting in domain administrator access within minutes.
As AI-powered attack frameworks like CAIRN and PentAGI become more prevalent, Google anticipates a decrease in human involvement in cyberattacks, along with increased stealth and sophistication.
Brief written by urgent.news; may contain errors — check the original.
Source: techradar.com