Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

cloud.google.com+1.cloud.google.com.siliconangle.com+1.Attackers have started running AI agents that scan networks, break in and move through them with little human direction, according to a report Google Threat Intelligence Group (GTIG) released Monday, November 16, 2026. The report, the latest edition of its quarterly AI Threat Tracker, says "the adversarial AI landscape has reached a critical inflection point." It also documents what GTIG calls the first case of threat actors using AI frameworks to "fully automate vulnerability discovery and exploitation without a human in the loop".cloud.google.com
GTIG found an exposed staging server running a multi-agent framework called exploit_agent. The framework connects to several frontier and local large language models. It compares the code of popular web applications before and after patch announcements, identifies likely flaws, writes exploits and runs them automatically. The server also held data stolen from victims, along with scripts for brute-forcing, installing webshells and removing webshells left by rival attackers. Telemetry suggested the operator was going after French media, entertainment and education websites. Russian-language comments in a configuration file told the user to enter API keys "right here, not in the chat".siliconangle.com+1
This is a change from two months ago, when GTIG said it had not seen fully autonomous attack pipelines used against real targets. Google says AI is shortening the time between a patch's release and a working exploit. In its count, total vulnerability disclosures and exploits nearly doubled between January and August 2026, and zero-day exploits jumped to 22 in August.cloud.google.com+1
GTIG and Mandiant say multiple threat actors are misusing open-source agentic penetration-testing tools, including ARTEX, CAIRN, CyberStrikeAI and PentAGI. One actor used CAIRN in intrusions against U.S. aerospace targets. Between late August and early October, another actor used a customized PentAGI platform to scan a wide range of organizations in Asia. Google also reported that a group with links to China was building its own automated attack framework, likely for espionage.cloud.google.com
The report follows other industry warnings. Check Point found more than 2,500 systems hosting ARTEX, and Arctic Wolf documented attackers running a fork of it against web services and cloud environments, according to Security Boulevard. GTIG's September tracker described an agent-driven credential-harvesting campaign that was planned and carried out in under six hours.securityboulevard.com+1
Google says the frameworks "currently trade stealth for velocity." Unrestricted autonomous execution generates forensic traffic and leaves code-level traces that make detection more likely. The company expects less human involvement and more stealth as the tools and models improve. Google, owned by Alphabet , said it disabled the accounts and assets linked to misuse of its Gemini models and used the findings to strengthen its classifiers.siliconangle.com+1