Researchers found exposed gang servers showing ransomware crews using commercial AI to plan attacks, and the tools they target are common in many offices.
Two ransomware gangs, INC and The Gentlemen, have been caught using commercial AI chatbots to plan attacks on companies. Security firm SOCRadar found seven attacker servers left open on the internet. They held thousands of AI chat messages and 46 custom ransomware programs, already built for six victims. Ransomware locks a company's files and demands payment to unlock them.
SOCRadar spotted the first open server on September 1. Login details in its logs led to six more. Inside were 8,257 messages across 133 AI sessions, dated from mid-July to September 30. The AI models included Claude Opus 4.6 and Kimi K3, reached through a service called OpenRouter.
The AI did not act alone. It planned reconnaissance, wrote commands, sorted stolen data and drafted extortion messages, but waited for a human to approve each step. The recorded spending was just over $23,920, while the gangs reportedly demand ransoms in the millions.
When a chatbot refused a request, the operator did not need a clever hack. He reused a template titled "Authorized Red Team Assessment Request", pretending to be a hired security tester. The AI cannot check who is typing, so the claim worked. One view in the coverage is that the fix has to come from vetting who opens accounts and watching usage patterns, not from reading the message.
The analysts counted 62 named organizations, 36 of them in the United States. Manufacturing and energy made up 16.1% of targets, and healthcare and pharma 14.5%. The gangs say they got in mostly through three tools: FortiClient EMS, SimpleHelp and cPanel/WHM. Those figures come from the criminals' own logs and are not independently verified.
Businesses that do not use those three tools are not among the entry points the gangs named. That is not a guarantee of safety, but it is a smaller risk.
The alert box above lays out the steps. In short, find out whether your business uses those tools, update them, make sure backups can be restored, and warn staff about extortion emails that may look polished and personal.
The report does not confirm whether the attack wave planned for October 1 went out. It also assigns low confidence to a formal alliance between the two gangs. It sees moderate to high confidence that they share an operating environment, possibly through one supplier of tools and infrastructure.
This is not the first such case. In August, Gambit Security reported a suspected Gentlemen affiliate using Claude Code against at least eight organizations. Anthropic has said it monitors abuse of its models in a September threat report.
Check whether your business runs FortiClient EMS, SimpleHelp or cPanel/WHM, and install the security updates for them now.
A viral ChatGPT prompt turns what the chatbot remembers about you into a room image. Posting it publicly could give scammers clues to target you.
Read more: Viral ChatGPT room image could hand scammers clues about youDARPA wants AI to hunt for design flaws in stock, bond, energy and cloud markets, so enemies cannot use them to cause economic damage that reaches households.
Read more: Pentagon wants AI to find weak spots in stock and energy marketsTuring Award winner Richard Sutton says more computing power alone will not make AI truly smart. Big AI companies disagree, which matters for anyone betting on AI.
Read more: AI pioneer says piling on computing power won't make machines smart
