Autonomous AI-driven worm can reason its way through corporate networks - Help Net Security

Summary

Researchers from University of Toronto, Vector Institute, and University of Cambridge developed an AI-driven worm that uses a large language model to autonomously reason through corporate networks, identify vulnerabilities, and create exploitation strategies on the fly. The prototype successfully identified and exploited vulnerabilities across diverse systems by running locally on compromised hardware, demonstrating the emerging threat of AI-powered autonomous cyberattacks.

Tags

ai-worm · cybersecurity · autonomous-attack · llm · vulnerability-exploitation

Key entities

Zeljka Zorz (person, 0.95) · University of Toronto (organization, 0.95) · Vector Institute (organization, 0.95) · University of Cambridge (organization, 0.95) · Help Net Security (organization, 0.95) · CleverHans Lab (organization, 0.9) · large language model (technology, 0.95) · AI-driven worm (technology, 0.95) · autonomous reasoning (concept, 0.9) · vulnerability exploitation (concept, 0.95) · Toronto (location, 0.9) · Cambridge (location, 0.9) · AI worm research study (event, 0.85)

Classification

analysis · language en · status final

Provenance

claude-haiku-4-5 via @stacklist/be@0.1.0, confidence 0.85, 24 Jun 2026

Full body

/api/public/card/71195774-537c-48b8-8f37-3329aeabe037.md