Researchers at the University of Toronto have identified a new type of computer worm that utilizes artificial intelligence to autonomously spread across networks. Unlike traditional worms that exploit specific software flaws, this AI-powered worm can assess vulnerabilities in real time and generate tailored attack strategies. In controlled tests, it successfully compromised nearly 75% of a simulated corporate network, which included various devices like Linux, Windows, and IoT devices, all without any human intervention. The worm operates using open-source large language models on infected devices, making it less susceptible to common cybersecurity measures. The economic implications are significant, as the worm effectively incurs no cost for spreading to additional machines by utilizing the computing resources of its victims. Researchers emphasize that this study is a warning about the potential for AI-enabled malware to evolve and adapt, urging governments and cybersecurity professionals to enhance their defenses against such emerging threats.