AI Stack Alert High TD-2026-0056 For SOC / IR, AppSec, AI Eng, CISO Unconfirmed AI-assisted
Claude document-processing flaw lets attackers run code via temporary file
Attackers can exploit a flaw in Claude's document-processing tool. The attack creates a temporary file that Claude can access and run on a user's machine. The source does not give affected versions, a CVE, or patch status.
Why it matters
Users of Claude's document-processing tool may have code run on their machines through a crafted temporary file. An attacker who succeeds could execute code under the user's session.
Do this
- Check Anthropic's official channels for an advisory, affected versions, and a fix for the document-processing tool, and apply any update when available.
- Restrict or disable Claude document-processing features on endpoints until a fix is confirmed, especially on Linux hosts.
- Hunt for unexpected child processes spawned by Claude or its helper processes, and for unfamiliar temporary files created in user temp directories.
- Review which local paths and files Claude-enabled tools can read and execute, and reduce those permissions to the minimum needed.
Signals
- NoCISA KEV
- 30/100Priority
- 40%Confidence
Evidence
“Attackers can exploit a flaw in Claude's document-processing tool by creating a temporary file that Anthropic's AI assistant can access and run on a user's machine.”
Sources
Affected / related
Claude, Anthropic
Published 2026-11-09 18:08 UTC · updated 2026-11-12 16:50 UTC · AI-assisted summary, checked against quoted sources · Unconfirmed: single non-official source · How we score