Google Warns the Era of Autonomous AI Attacks Has Arrived
Google said the system cut patch-to-exploit time to under three hours and enabled credential theft without human direction.
- According to a new AI Threat Tracker report published by the Google Threat Intelligence Group and Mandiant, threat actors have crossed a major operational threshold by deploying autonomous AI agents to execute end-to-end cyberattacks without human intervention.
- Google researchers documented what they assess to be the first fully automated AI-driven cyberattack—a system dubbed "exploit_agent" linked to a Russian threat actor—which independently handles vulnerability discovery, exploitation, and post-compromise activity.
- The report warns that malicious actors have rapidly evolved beyond basic text prompting to utilizing multi-agent frameworks, including open-source offensive tools like CyberStrikeAI, HexStrike AI, and PentAGI, to connect large language models directly to hundreds of exploitation utilities.
- Google projects that simple attacks executed by multiple AI agents with minimal human oversight could emerge within the next three months, with more sophisticated, fully autonomous agentic campaigns expected within three to six months.
- In response to the accelerating pace of automated exploitation—which saw overall vulnerability disclosures nearly double between January and August 2026—Google stated it has disabled identified malicious assets, strengthened Gemini safety classifiers, and released new defensive hardening guidelines.
107 Articles
107 Articles
Russian AI bots pull off first fully automated cyber attack — and Google caught them
Google’s cyber threat intelligence division identified hackers on Monday who used AI to “fully automate vulnerability discovery and exploitation without a human in the loop for the first time.”
NEW: Google Issues Ominous Warning Amid String Of Cyber Outages
Google announced this week that artificial intelligence has reached a point where computer programs can find security weaknesses in software and try to break in on their own, with little or no human help at each step. The company’s Threat Intelligence Group plainly stated in a new report that “the era of autonomous attacks has arrived.” Researchers identified a tool — identified as exploit_agent by those operating it — sitting on an unprotected …
Google Warns Fully Autonomous AI-Powered Cyberattacks Could Soon Become Common
Google Threat Intelligence Group researchers said attackers are moving beyond manual interactions with large language models toward autonomous, multi-agent frameworks capable of executing complex intrusions at greater speed.
Google announced the first cyber attack in history in which artificial intelligence itself sought vulnerability and hacked into software.
Google has for the first time caught cybercriminals who use artificial intelligence to attack without anyone by means: the AI finds the security failure and exploits it alone. It is told by the AI Threat Tracker report of the Google Threat Intelligence Group (GTIG), published on November 16, 2026. It signs it together ... Continue reading "The age of autonomous attacks with AI has arrived: Google catches for the first time hackers who find and e…
Coverage Details
Bias Distribution
- 39% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium

































