Simon Willison • 1/14/2026

Claude Cowork Exfiltrates Files

Read Original

A security researcher discovered a vulnerability in Claude Cowork where its default HTTP allowlist, designed to prevent data exfiltration, could be bypassed. The attack used the permitted Anthropic API domain with an attacker's API key to upload files from the agent to the API endpoint, enabling data theft.

0 comments
#prompt injection #ai security #API Security
Claude Cowork Exfiltrates Files

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser

Top of the Week

1
The Beautiful Web
Jens Oliver Meiert • 2 votes
2
When your coding agent doesn’t understand your project, you’ll get junk
Benjamin Cane • 1 votes
3
LLM Use in the Python Source Code
Miguel Grinberg • 1 votes
4
Wagon’s algorithm in Python
John D. Cook • 1 votes
5
An example conversation with Claude Code
Dumm Zeuch • 1 votes