AI Agents Enable Adaptive Computer Worms
2026-06-02 • Cryptography and Security
Cryptography and SecurityArtificial IntelligenceMachine Learning
AI summaryⓘ
The authors studied a new type of computer worm that uses artificial intelligence to create custom attack methods for each target it finds. Unlike traditional worms that rely on fixed vulnerabilities, this AI-powered worm uses infected machines to run large language models, helping it think and adapt its attacks in real time. The worm spreads across different devices like Windows, Linux, and IoT systems by exploiting common network weaknesses, costing attackers almost nothing per new infection. The authors highlight that such self-sustaining AI malware can work without human control and traditional security measures may not stop it.
computer wormmalwareartificial intelligencelarge language modelsnetwork securityIoT devicescyber-threatsvulnerabilitiesautonomous systems
Authors
Jonas Guan, Tom Blanchard, Hanna Foerster, Hengrui Jia, Gabriel Huang, Nicolas Papernot
Abstract
A computer worm is malware that spreads on a network by replicating itself from one machine to another. Traditional worms, like WannaCry, exploited predetermined vulnerabilities, and their spread can be halted by patching those vulnerabilities. Here we show that artificial intelligence (AI) agents enable a fundamentally new threat: a worm that generates tailored attack strategies to each target it encounters. The worm parasitically uses compromised machines to run open-weight large language models (LLMs) to sustain its reasoning, or extend its reach for further attacks. Deployed on a network of machines spanning Linux, Windows, and IoT (Internet of Things) devices, the worm propagated by exploiting common, real-world corporate network vulnerabilities. Since the worm is powered by stolen compute, the attacker's marginal cost per new infection is zero. This creates a destabilizing economic asymmetry between attackers and defenders. Moreover, because the worm requires no commercial AI platform, centralized safety controls, such as service refusals or rate limiting, are structurally irrelevant. Our results demonstrate that self-sustaining AI-driven cyber-threats are no longer theoretical. We must prepare for autonomous generative adversaries: malware systems that propagate without human operators and are defined not by fixed exploit code, but by the capacity to reason about targets, adapt to observations, and synthesize attack logic in real time.