Researchers Build Self-Replicating AI Worm That Operates Entirely on Local, Open-Weight Models
Researchers at the University of Toronto have developed a proof-of-concept computer worm that leverages a locally hosted open-weight large language model to autonomously reason through a network, craft specific attack strategies for each target, and self-replicate. The worm operates entirely without human intervention and does not rely on any commercial AI service. The team posted a preprint describing their work to arXiv. This demonstrates a new capability for AI-driven malware that can adapt its behavior in real time using local models, potentially lowering the barrier for sophisticated automated attacks.
// why it matters
Self-replicating AI worms using local models could enable autonomous, adaptive malware that evades cloud-based detection.