AI news ·
Researchers build AI-powered computer worm that learns as it spreads between devices
Researchers built a working AI-powered computer worm that spreads between devices and adapts its attacks autonomously. Experts warn current defenses are unprepared for malware that learns as it spreads.
Researchers Build AI-Powered Computer Worm That Learns as It Spreads
Scientists have created a working prototype of malware that uses artificial intelligence to spread between devices autonomously and adapt its attacks-demonstrating a cybersecurity threat that experts say the world is unprepared to defend against.
The researchers built the worm using freely available AI models, not proprietary systems from companies like OpenAI or Anthropic. The prototype was created in an isolated virtual environment and has not been peer-reviewed, but the findings were posted on arXiv.org.
Why This Matters
Traditional computer worms follow fixed instructions programmed by their designers. AI-powered worms can learn and adapt as they spread, identifying and exploiting vulnerabilities that weren't anticipated by system designers.
David Lie, a professor at the University of Toronto who reviewed the research, called it a "wake-up call" for cybersecurity experts. "The demonstration here is that there's a motivation to do this sooner rather than later," he said.
The implications extend across critical infrastructure. Water systems, power grids, financial networks, healthcare systems, and government operations all depend on networked computers that could be targeted by AI-powered malware.
The Defense Side
Lie noted that AI is a dual-use technology. The same capabilities that enable adaptive attacks can be used to build defenses. "They're mirrors of each other," he said.
The research underscores why cybersecurity professionals need to understand both the offensive and defensive applications of AI. Those working in the field should consider exploring AI for Cybersecurity Analysts and AI Research Courses to stay current with emerging threats and countermeasures.
The researchers emphasized that their work is intended to prompt faster development of defenses, not to enable attacks. The prototype's creation in a controlled environment demonstrates the threat without releasing functional malware into the wild.