Reddit
Giving Claude Full macOS Access Sparks 2.4K-Upvote Debate on Risk vs Power
A screenshot of Claude with full macOS Computer Use access — controlling Finder, Terminal, and system settings — hit 2,424 upvotes and 75 comments on r/ClaudeAI. The post ('Good idea or nah?') sparked a security debate about granting AI agents unrestricted OS access. Comments split between power users describing productivity gains from full-system Claude access and security-minded practitioners warning about unintended file modifications and data exposure. This is a leading indicator of how developer trust boundaries are shifting as agentic AI becomes a daily tool.
Source
↳ Follow the thread