Windows CLOSEDQUORUM malware uses AI models to autonomously select post-compromise actions

A new Windows malware called CLOSEDQUORUM can query up to four LLM providers - Google Gemini, DeepSeek, Qwen, and Mistral - to autonomously select from predefined post-compromise actions, including stealing users’ credentials and cryptocurrency wallets. Once deployed, the malware does not require continued commands from a human operator, according to Cisco Talos, which describes it as, to its knowledge, the first publicly documented Windows implant to use this approach for command-and-control (C