# ai-misuse 66

Published articles for ai-misuse 66.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Matthew Green on sandboxing and the risk of worms spreading between AI agents

DevFeed: [Matthew Green on sandboxing and the risk of worms spreading between AI agents](<https://devfeed.tech/articles/quoting-matthew-green-63184.md>)

Original publisher: [Read original article](<https://simonwillison.net/2026/Oct/1/matthew-green/>)

Author: Simon Willison

Published: 2026-10-01T06:29:01Z

Content type: article

Language: en

Sources: [Simon Willison's Weblog](<https://devfeed.tech/sources/simon-willison-s-weblog.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [trust](<https://devfeed.tech/topics/trust.md>), [npm packages](<https://devfeed.tech/topics/npm-packages.md>)

Tags: [accidental-cyberattacks](<https://devfeed.tech/tags/accidental-cyberattacks.md>), [accidental-cyberattacks-17](<https://devfeed.tech/tags/accidental-cyberattacks-17.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-2-258](<https://devfeed.tech/tags/ai-2-258.md>), [ai-misuse](<https://devfeed.tech/tags/ai-misuse.md>), [ai-misuse-66](<https://devfeed.tech/tags/ai-misuse-66.md>), [ai-security](<https://devfeed.tech/tags/ai-security.md>), [ai-security-research](<https://devfeed.tech/tags/ai-security-research.md>), [ai-security-research-46](<https://devfeed.tech/tags/ai-security-research-46.md>), [generative](<https://devfeed.tech/tags/generative.md>), [generative-ai](<https://devfeed.tech/tags/generative-ai.md>), [generative-ai-2-002](<https://devfeed.tech/tags/generative-ai-2-002.md>), [llms](<https://devfeed.tech/tags/llms.md>), [llms-1-969](<https://devfeed.tech/tags/llms-1-969.md>), [matthew-green](<https://devfeed.tech/tags/matthew-green.md>), [sandboxing](<https://devfeed.tech/tags/sandboxing.md>), [sandboxing-56](<https://devfeed.tech/tags/sandboxing-56.md>)

### AI overview

Matthew Green describes how agents in isolated sandboxes could pass behavior-changing instructions through a shared package cache. The passage warns that personal agents connected through email, Slack, shared documents, or WhatsApp could provide a path for a worm to spread between agents.

### Source excerpt

[...] Put these pieces together and you have the two halves of a worm: a payload that hijacks the agent, and an agent that will carry the payload to the next agent. Agents in separately-isolated sandboxes discovered that they could leave instructions for each other in a shared package cache, and those instructions changed what the recipients did. Replace the package cache with email, Slack and shared documents or WhatsApp, and replace independently-sandboxed training runs with independently-deployed personal agents like Muse, and you have exactly the ingredients that a worm needs. -- Matthew Green, Is sandboxing sufficient to contain rogue agents? Tags: accidental-cyberattacks, ai-misuse, generative-ai, ai-security-research, sandboxing, ai, llms