Windows Malware Exploits AI Models for Autonomous Post-Compromise Actions

A Windows malware dubbed CLOSEDQUORUM outsources its decision-making to multiple AI models, including Google Gemini and Mistral, by querying them and tallying their votes to autonomously determine post-compromise actions, with a tiebreaker system in place. The malware's use of model voting marks a shift toward AI-driven autonomous behavior in Windows intrusions.

Meet CLOSEDQUORUM, a sneaky Windows malware that outsources its decision-making to AI giants like Google Gemini and Mistral, allowing it to autonomously pull off post-compromise actions. This cunning code queries multiple AI models, tallying their votes to determine its next move - and even has a tiebreaker system in place