Researchers Build AI-Powered Worm With Public LLM

A team from University of Toronto, University of Cambridge, and elsewhere built an AI-powered worm using an open source LLM. The worm spreads autonomously across devices in a network with common vulnerabilities like reused passwords. It generates tailored attacks for each target and operates in a controlled environment with Linux, Windows, and IoT devices. The threat predates large-scale LLMs and demonstrates real-world feasibility of autonomous malware.