University of Toronto researchers have developed the first autonomous, AI-powered computer worm capable of learning and adapting its attack strategies in real time to compromise network-connected devices.
Key Points
- The proof-of-concept malware utilizes publicly available large language models to identify and exploit vulnerabilities without human intervention.
- Unlike traditional static malware, this autonomous worm can spread across networks independently and repurpose infected device processing power to sustain its operations.
- Researchers developed the tool in a controlled academic setting to proactively identify security risks before malicious actors can weaponize similar AI technologies.
- Experts recommend securing devices by changing default passwords, utilizing passkeys, enabling multi-factor authentication, and installing software updates immediately.
- The study highlights that smart home appliances, routers, and cameras are particularly susceptible to these adaptive threats if not properly isolated on guest networks.