Highwise Tech
World

Cisco Talos Released Framework to Track AI Malware

The new CAIRN framework provides a standard for identifying and classifying autonomous, LLM-integrated hacking tools.

Updated on Sept. 22, 2026 in Artificial Intelligence

Bold flat-color editorial illustration showing a central geometric structure connected to smaller floating blocks, representing AI malware classification.
Cisco Talos has launched the CAIRN framework, an open-source tool designed to track and categorize autonomous, AI-integrated malware and agentic cyber threats. AI Illustration. Upload story photo >

Live Poll

Do you believe the rise of AI-powered malware makes your personal digital information less secure?

Cisco Talos has released the open-source CAIRN framework to track the rise of AI-integrated malware. Researchers utilized the tool to identify CLOSEDQUORUM, an autonomous malware strain that operates without human input by querying multiple large language models.

Why it matters

The framework addresses the shift toward agentic AI in cyberattacks, where malicious tools leverage autonomous command systems to bypass traditional manual oversight. This development marks an effort to categorize and mitigate threats that use AI to optimize credential theft and financial fraud.

The CAIRN framework analyzes artifact metadata to tag AI-integration characteristics, while the CLOSEDQUORUM malware polls four LLMs—DeepSeek, Qwen, Mistral, and Google Gemini—to reach consensus on its actions.

The players

Cisco Talos

A division of Cisco Systems that provides threat intelligence, vulnerability research, and security incident response services.

CERT-UA

The Computer Emergency Response Team of Ukraine, which monitors and defends critical national infrastructure against cyber threats.

The details

CAIRN, or Classification and AI-integrated Repository Network, functions by assigning unique IDs to malware samples based on evidence of LLM integration. The CLOSEDQUORUM malware uses this architecture to create a hive mind; by polling four independent models, the tool ensures high-availability commands for malicious tasks like credential harvesting, even if a single model provider becomes inaccessible or refuses a request.

Timeline

  1. July 2025: CERT-UA detected a phishing campaign utilizing LAMEHUG malware.

  2. 2025: CLOSEDQUORUM malware was first identified in connection with credit card fraud forums.

  3. September 21, 2026: Cisco Talos released the CAIRN framework.

The Tech Race

The introduction of CAIRN shifts the security industry from reactive identification toward systematic tracking of agentic malware. This development marks an extension of the detection standards established by the 2025 LAMEHUG phishing campaign to account for multi-model, autonomous architectures.

Organizations and security researchers can now implement the CAIRN framework to audit systems for the specific metadata signatures of autonomous AI-based malware. This tool is intended for technical security teams to prioritize identifying threats that no longer rely on traditional human-command infrastructure.

The takeaway

The transition to agentic, autonomous malware signifies a new defensive challenge that requires automated, model-aware classification tools. Security professionals should monitor future updates to the CAIRN repository to track how these multi-model threat actors evolve in response to updated detection methods.

Further reading

For broader context on how autonomous systems are shaping security, see Artificial Intelligence.

Live Poll

Do you believe the rise of AI-powered malware makes your personal digital information less secure?