Hermes Vector

AI & SITrending

Google Detects First AI-Automated Cyberattack Without Human Intervention

Google Threat Intelligence Group says it identified an AI agent that autonomously finds and exploits software vulnerabilities, with Russian-language comments found in its configuration file.

By Hermes Vector

Checked against 3 sourcescloud.google.comGoogleupi.com
Illustration for: Google Detects First AI-Automated Cyberattack Without Human Intervention

In brief

  • GTIG identified exploit_agent, an AI framework that autonomously finds and exploits software vulnerabilities without human intervention.
  • The framework's configuration file contained Russian-language comments, and it targeted French media, entertainment, and education websites.
  • GTIG also observed threat actors using open source agentic frameworks including CAIRN, PentAGI, CyberStrikeAI, and HexStrike AI in live intrusions.
  • Zero-day exploits spiked to 22 in August 2026, up from a 2025 average of 8 per month.

Google Detects First AI-Automated Cyberattack Without Human Intervention

Google Threat Intelligence Group (GTIG) has identified an AI framework that autonomously searches for software vulnerabilities and exploits them without human involvement, according to a report published November 16, 2026.

The exploit_agent framework

GTIG identified an exposed staging server hosting a multi-agent vulnerability research and weaponization framework dubbed exploit_agent. The framework integrates with multiple frontier and local LLMs and is set up to conduct continuous differential analysis of popular web application software. It compares code before and after patch announcements, analyzes findings to identify potential vulnerabilities, drafts exploits, and automatically executes them without human intervention.

The server contained scripts capable of facilitating post-compromise activities, including internal reconnaissance, brute-forcing, installing webshells, and removing other attackers’ webshells. Victim telemetry on the server suggests the operator was specifically interested in targeting French websites associated with the media and entertainment industries, as well as the education sector.

The configuration file included Russian-language comments. One comment, translated to English, reads: "LLM API keys for auto-exploitation (exploit_agent pilot)." Another reads: "Enter the keys yourself — right here, not in the chat. Both lines are required."

Broader agentic threat activity

GTIG and Mandiant have observed multiple threat actors abusing open source, agentic penetration-testing frameworks to automate discrete phases and specific operational tasks during live intrusions. Frameworks observed include ARTEX, CAIRN, CyberStrikeAI, PentAGI, and HexStrike AI.

GTIG has independently observed open source tools CAIRN and PentAGI deployed in the wild, and has seen experimentation with CyberStrikeAI and HexStrike AI. CAIRN was observed used in intrusion activity against targets including in the aerospace sector in the US. Between late August and early October 2026, GTIG tracked a threat actor using a customized attack orchestration platform built off of the open source version of PentAGI used to conduct systematic reconnaissance and vulnerability scanning against a wide range of organizations in Asia.

Threat actors observed using these frameworks include financially motivated actors as well as suspected espionage threat actors with a PRC nexus. This includes evidence that a PRC-nexus group was developing its own automated attack framework for likely use in cyber espionage operations.

Vulnerability exploitation trends

In September 2026, GTIG analyzed vulnerability statistics and found that overall disclosures and exploits nearly doubled between January and August 2026. Zero-day exploits rose only marginally from a 2025 average of 8 per month to 11 per month in 2026, before spiking to 22 in the month of August.

In late August 2026, threat actors exploited two vulnerabilities in PaperCut print management servers as zero-days that, when chained together, allow for unauthenticated remote code execution: CVE-2026-81578, an authentication bypass flaw; and CVE-2026-82078, an unsafe dynamic class loading vulnerability. Using LLMs, security researchers and the cybersecurity company Huntress generated working exploits for the vulnerabilities within a day of disclosure. Greynoise reported a campaign that took place four days after the initial disclosure in which a threat actor likely used LLMs to build and test exploits for these vulnerabilities, assemble a target list, and execute a widespread exploitation campaign against hundreds of organizations.

Why it matters

GTIG states that threat actor adoption of these frameworks represents a paradigm shift towards intrusions occurring at machine speed. John Hultquist, Chief Analyst at Google Threat Intelligence Group, said: "For the most part, attackers aren’t using new groundbreaking methods to break into systems — they are still going after known, unfixed issues, just with much greater speed and scale. Instead of relying on manual effort to keep up, we will have to embrace agentic defense. Or we risk falling behind."


Sources

This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.

Questions readers ask

What is the exploit_agent framework identified by Google?
The exploit_agent framework is a multi-agent vulnerability research and weaponization tool that autonomously searches for software vulnerabilities and exploits them without human involvement. It integrates with multiple LLMs to conduct continuous differential analysis of web application software and automatically executes drafted exploits.
When was the report about the AI-automated cyberattack published?
The report was published on November 16, 2026.
Which specific sectors were targeted by the operator of the exploit_agent server?
Victim telemetry suggests the operator was specifically interested in targeting French websites associated with the media and entertainment industries, as well as the education sector.
What other agentic frameworks have been observed in the wild by GTIG and Mandiant?
Frameworks observed include ARTEX, CAIRN, CyberStrikeAI, PentAGI, and HexStrike AI. GTIG has independently observed CAIRN and PentAGI deployed in the wild, and has seen experimentation with CyberStrikeAI and HexStrike AI.
How did zero-day exploit statistics change in 2026 compared to 2025?
Zero-day exploits rose marginally from a 2025 average of 8 per month to 11 per month in 2026, before spiking to 22 in the month of August.

The day's checked news, by email

One short email a day with the top stories. Free, and you can unsubscribe in one click.

Or follow for the latest news:

How do you feel about this story?

Comments 0

No comments yet. Start the discussion.

AI & SI

OpenAI and Chile sign AI adoption partnership

OpenAI and Chile's Ministry of Science, Technology, Knowledge and Innovation signed a memorandum of understanding to support responsible AI adoption in public administration, civil servant training, and local startup development.

Checked against 1 sourceOpenAI