Story perspectives

AI-Built Worm Autonomously Exploits Flaws, Sparks Cyber Panic

6/3/2026

46 8

Summary

Story summary

University of Toronto researchers created an AI-generated worm that can exploit any known software flaw and spread autonomously.
The prototype moved across an isolated test network without human intervention while the network stayed offline.
Anthropic restricted Claude Mythos to roughly 40 critical-infrastructure groups, fearing it could let hackers exploit vulnerabilities faster.
The researchers warned AI-driven worms could launch a new, hard-to-defend hacking era.

1 of 1

Story summary

University of Toronto researchers created an AI-generated worm that can exploit any known software flaw and spread autonomously.
The prototype moved across an isolated test network without human intervention while the network stayed offline.
Anthropic restricted Claude Mythos to roughly 40 critical-infrastructure groups, fearing it could let hackers exploit vulnerabilities faster.
The researchers warned AI-driven worms could launch a new, hard-to-defend hacking era.