University of Toronto researchers demonstrated how artificial intelligence can create dangerous computer worms capable of targeting any known vulnerability. The prototype spread across a test network without human intervention, raising concerns about a new era of AI-powered cyber threats that could spread faster than traditional malware.

The research adds to growing evidence that AI advances are creating unprecedented cybersecurity risks. Companies like Anthropic and OpenAI have already limited releases of powerful technologies due to security concerns. The Toronto team used open-source AI, meaning restrictions on its use are impossible, unlike proprietary systems from major tech firms.

The technology presents significant challenges for businesses, requiring more robust patching strategies. Security experts warn that while current AI systems may be prone to errors, their capabilities will improve. The research underscores an emerging market for AI-powered defense tools, as companies race to develop countermeasures against these evolving threats.