Researchers Prototype AI-Powered Internet Worm with Self-Executing LLM

Researchers have prototyped an AI-powered internet worm that carries and executes its own large language model (LLM) on compromised computers. The development aligns with John Brunner’s 1975 conceptualization of a self-replicating computer worm from The Shockwave Rider. No specific dates, technical identifiers (e.g., CVE IDs), or affected systems were disclosed in the report. The worm’s capability to deploy an LLM on breached machines represents a novel propagation method. The prototype was documented in a research publication linked from cleverhans.io and referenced in a New York Times article dated June 2, 2026. No direct impacts beyond the proof-of-concept demonstration were described.