Reddit users question Claude Cowork file access: what stops it reading beyond the granted folder?
OriginalPosition1 · reddit · 2026-11-15
A Reddit user questions how much to trust Claude Cowork with local files: what technically prevents the agent from accessing files outside the granted folder on a Mac, and whether there's any way to audit which files it actually touched, or if users must simply trust the built-in permissions. The thread reflects broader concerns about sandbox boundaries and observability for desktop AI agents.
More from Safety
- Team builds formal verification system to monitor and control autonomous AI agents — Smooth_Profit4543 · 2026-11-15
- Jack Clark urges Congress to crack down on chip smuggling, cites $390M Singapore scheme to transfer servers to DeepSeek — basedjensen · 2026-11-15
- The Confused Deputy Problem: when high-privilege processes are tricked into reading low-privilege files — wavefnx · 2026-11-15
- Chinese Firms Smuggle Nvidia Chips While Beijing Bans Them — Hawk Consensus — theteknosaur · 2026-11-15
- Ramez: Highest-Benefit AI Policy Is Forcing Labs to Disclose Research Data — latticecut · 2026-11-15
- rea + mitmproxy: a 60-second workflow that makes Claude Code expose what your apps are phoning home — Kyrannio · 2026-11-15