Cisco Talos Releases AI Malware Tracking Tool, Uncovers Autonomous Hacking System Guided by Four LLMs

Cisco Talos researchers released an open-source framework in September 2026 designed to identify and classify malware that incorporates artificial intelligence — and it has already surfaced a previously undocumented hacking tool that operates without any human involvement.

The framework, called Cognitive Artifact Intelligence Research Network (CAIRN), works by detecting the digital fingerprints that AI integration leaves behind in malware. It tags samples with unique identifiers, groups them by shared traits, and surfaces potential trends across its library. Using CAIRN, researchers identified a Windows-based hacking tool they have named CLOSEDQUORUM, which polls up to four large language models — DeepSeek, Qwen, Mistral, and Google Gemini — to determine its next moves inside a target system. The malware draws on whichever services are available, creating enough redundancy that no human input is required at any point.

CLOSEDQUORUM is designed to steal login credentials and cryptocurrency. Cisco Talos researchers found links between the tool and cybercriminal forums discussing credit card fraud dating to 2025, though they could not confirm who developed it or whether it has been deployed in real-world attacks.

Ryan Fetterman, the Cisco Talos security researcher who led CAIRN’s development, said the project grew partly out of surprise. After a July 2025 warning from Ukraine’s CERT-UA about phishing malware communicating with an LLM via a Hugging Face API, Fetterman expected a surge in AI-enabled malware. A retrospective this past summer turned up only about nine named malware families — some of them research proofs of concept. CAIRN has since helped him identify roughly 20 additional examples.

“While I do think this is still largely experimental for attackers, the landscape is a lot more complex and diverse than has been publicly reported,” Fetterman said.

Matt Olney, senior director of threat intelligence at Cisco Talos, said AI is shifting from a productivity aid to an operational tool for attackers, potentially allowing them to run more campaigns across more systems with less direct oversight. CAIRN is being released publicly in the hope that the broader security community will adopt it to monitor how this threat evolves.