Skip to main content
World Today News
  • Home
  • News
  • World
  • Sport
  • Entertainment
  • Business
  • Health
  • Technology
Menu
  • Home
  • News
  • World
  • Sport
  • Entertainment
  • Business
  • Health
  • Technology
Google Identifies First Fully Autonomous AI Cyberattack via Exploit_agent System

Google Identifies First Fully Autonomous AI Cyberattack via Exploit_agent System

November 22, 2026 Technology Desk Technology

Google analysts have identified the first cyberattack carried out entirely by an artificial intelligence system, moving from human-led operations to machine-speed intrusions.

The discovery, detailed in a report from Google’s Threat Intelligence Group and Mandiant, involves an autonomous framework capable of discovering software vulnerabilities and launching exploits without human intervention.

Researchers uncovered an unsecured staging server hosting a system named exploit_agent, which connects to multiple advanced and locally hosted language models, TechRadar reported. The framework continuously analyzes software code to identify differences before and after security patches are released. Once a flaw is detected, the system automatically drafts and executes an exploit.

The operation shows a major change in how cybercriminals use AI, according to CyberScoop.

While earlier attacks relied on human-supervised chatbots to draft code or phishing lures, this new approach allows for the automation of internal network mapping, password guessing, and the planting of webshells.

“While the mechanics of the intrusions remain the same, such as using compromised credentials and targeting unpatched systems, the speed and scale at which frameworks can exploit these known weaknesses introduces a new defensive challenge,” Google analysts stated, as noted by news.yahoo.com.

The shift toward autonomous frameworks has coincided with a sharp rise in zero-day exploits. Google recorded 22 zero-day vulnerabilities in August 2026, compared to an average of eight per month in 2025 and 11 per month earlier this year, according to CyberScoop.

Time Period Average Zero-Day Exploits Per Month
2025 8
Early 2026 11
August 2026 22

Scripts on the same server also handled internal reconnaissance, password guessing, and the installation and removal of webshells, including one that erased webshells planted by rival attackers, according to the quarterly report. Researchers stopped short of naming a group or state, basing attribution on Russian-language comments in the code and a configuration file instructing users to enter API keys for “auto-exploitation” there rather than in a chat, CyberScoop reported.

Google analysts said that while these frameworks compress attack timelines to machine speed and achieve operational objectives, they currently trade stealth for velocity, and the company anticipates reduced human involvement along with increased stealth and sophistication in future activity, TechRadar reported.

Google’s Threat Intelligence Group now publishes major findings jointly with Mandiant, a Google Cloud subsidiary since September 2022, as seen in the exploit_agent report that details the first fully AI-driven cyberattack pipeline. This marks a shift from Mandiant’s earlier independent work, such as its February 2013 report attributing espionage to China’s PLA Unit 61398 while operating as a standalone firm.

No Human Hacker: Google Catches a Fully AI-Run Cyberattack — Headline Roots

Attribution of the exploit_agent remains preliminary. Investigators found Russian-language comments within the system’s configuration files, and victim data suggests the operator targeted media, entertainment, and education websites in France, business-standard.com reported.

The security community has also observed the use of open-source offensive frameworks like CAIRN and PentAGI.

Google confirmed that a customized PentAGI platform was used to probe organizations across Asia for vulnerabilities between late August and early October 2026.

“While these frameworks are capable of significantly compressing attack timelines to machine speed and achieving operational objectives, they currently trade stealth for velocity,” the Google report said. Researchers anticipate that future iterations will likely focus on increasing stealth to avoid detection by security teams.

Instagram

Share this:

  • Share on Facebook (Opens in new window) Facebook
  • Share on X (Opens in new window) X

Worth a look

  • Apple Intelligence Uses 9-16GB on iOS 27; Siri Language Workaround May Reclaim Space
  • Sortir à Paris: GTA 6 Locks Cheats Behind Story Completion, Lists 61 Codes
  • Apple Plans Autonomous AI Agent Launch by 2027 (time.news)

Related

Search:

World Today News

World Today News is your trusted source for global journalism — breaking headlines, in-depth analysis, and reporting from around the world.

Quick Links

  • Privacy Policy
  • About Us
  • Accessibility statement
  • California Privacy Notice (CCPA/CPRA)
  • Contact
  • Cookie Policy
  • Disclaimer
  • DMCA Policy
  • Do not sell my info
  • EDITORIAL TEAM
  • Terms & Conditions

Browse by Location

  • GB
  • NZ
  • US

Connect With Us

© 2026 World Today News. All rights reserved. Your trusted global news source directory.
For contact, advertising, copyright, issues email: [email protected]

Privacy Policy Terms of Service