Brief

Researchers have developed a prototype AI-powered computer worm that can adapt its attack strategies in real-time to exploit specific vulnerabilities on infected machines. This new type of malware leverages open-weight large language models running on compromised systems, effectively eliminating the attacker's cost per infection and creating a significant economic imbalance for defenders. The worm's ability to learn and adapt makes traditional patching methods insufficient, highlighting an urgent need for new cybersecurity defenses against autonomous, generative threats. AI

Researchers at the University of Toronto have developed a new type of AI-powered cyberworm that can adapt its strategy as it spreads across networks. This worm, built using accessible AI models, can seize control of devices and networks at a low cost, posing a significant threat to various online systems. The team's findings, conducted in a secure lab, highlight the potential for misuse of open-weight AI models and raise concerns about current cybersecurity defenses. AI