Lawrence Jengar
Mar 23, 2026 22:21
Claude AI can now management your Mac straight – opening apps, navigating browsers, and operating dev instruments. Obtainable in analysis preview for Professional and Max subscribers.
Anthropic simply gave Claude the keys to your pc. The AI firm introduced Monday that Claude can now straight management macOS machines – clicking, scrolling, and navigating purposes to finish duties whereas customers step away.
The characteristic, launching in analysis preview for Claude Professional and Max subscribers, pairs with Dispatch, a cell device launched final week that lets customers assign duties to Claude from their telephones.
What Claude Can Truly Do
When Claude lacks a direct integration with a service, it falls again to controlling your display like a human would. The AI opens information, operates browsers, and runs improvement instruments routinely. Want a morning briefing compiled whereas commuting? Claude can pull emails and metrics utilizing your precise purposes.
For builders, the implications are extra concrete. Claude could make modifications in your IDE, run checks, and submit pull requests – all when you deal with different work.
Anthropic constructed the system with a permission-first method. Claude requests entry earlier than touching new purposes, and customers can halt operations at any level. The corporate additionally applied computerized scanning to detect immediate injection makes an attempt, a standard assault vector for AI methods with pc entry.
The Limitations Are Actual
Anthropic is not overselling this. “Pc use continues to be early in comparison with Claude’s potential to code or work together with textual content,” the corporate acknowledged. Complicated duties typically require a number of makes an attempt, and screen-based management runs slower than direct API integrations.
Some purposes stay off-limits by default. Anthropic recommends avoiding delicate knowledge and sticking to trusted apps in the course of the analysis preview. The corporate printed security tips however famous that “threats are consistently evolving.”
Broader Context
This launch follows a busy interval for Anthropic. The corporate shipped Claude Opus 4.6 on February 5 and Claude Sonnet 4.6 on February 17, each focusing on complicated agentic workflows – precisely the type of duties that profit from pc management.
The timing additionally comes after Anthropic’s February conflict with US federal businesses. The corporate refused to take away contractual prohibitions in opposition to utilizing Claude for mass surveillance and autonomous weapons, resulting in a phase-out of presidency utilization.
Pc management represents the following frontier in AI assistants – shifting from dialog companions to autonomous brokers that function in the true world. OpenAI and Google have signaled related ambitions, however Anthropic shipped first on desktop.
For now, the characteristic requires macOS with the desktop app operating. Cellular customers can assign duties by Dispatch, however the precise pc management occurs on the Mac. Anthropic plans to study from early utilization earlier than increasing platform assist.
Picture supply: Shutterstock
