Contents
Key Highlights
- Anthropic introduced remote Mac computer control capabilities for Claude AI
- Claude executes tasks by operating applications, navigating websites, and managing documents
- The capability integrates with Dispatch, enabling task delegation through mobile devices
- The company cautions users that the technology remains experimental and recommends caution with sensitive information
- This launch follows OpenClaw’s viral success, intensifying competition in the agentic AI sector
Anthropoic has unveiled a groundbreaking capability that enables Claude AI to remotely operate users’ computers for task completion. The company made the announcement on Monday, with immediate availability for Claude Pro and Claude Max paying members.
The functionality is currently exclusive to MacOS systems. Users issue commands through their mobile devices, prompting Claude to execute operations on their desktop machines — launching applications, conducting web searches, populating data sheets, or organizing documents.
A demonstration video from Anthropic featured a scenario where someone running behind schedule requested Claude to convert a presentation into PDF format and include it in a calendar appointment. Claude executed the entire workflow autonomously.
The system interfaces with platforms such as Google Calendar and Slack using integrated connections. When native connectors are unavailable, Claude employs keyboard and mouse controls to navigate applications manually.
Claude requires explicit authorization before interacting with any new application. Users maintain the ability to terminate operations at any point during execution.
Integration With Dispatch Functionality
This computer control feature complements Dispatch, a companion tool Anthropic introduced the previous week within Claude Cowork. Dispatch facilitates ongoing conversations with Claude across multiple devices while delegating extended assignments.
When combined, these capabilities enable automated functions including morning summary reports, software testing procedures, or inbox management.
Anthropoic acknowledges that sophisticated operations may require multiple attempts. The organization characterizes this release as an experimental preview designed to collect user insights and pinpoint areas requiring enhancement.
Privacy Concerns and Safety Measures
Anthropoic has implemented protective mechanisms to minimize vulnerabilities, featuring automated detection systems for prompt injection exploits. However, the company advises users against deploying the feature with applications containing confidential information — certain sensitive applications are restricted by default.
Cybersecurity professionals have raised wider concerns regarding agentic artificial intelligence. These systems can execute significant operations rapidly with minimal oversight. Additional risks include malicious users commandeering the agent to infiltrate private files or infrastructure.
Anthropoic committed to continuously enhancing security protocols as new vulnerabilities are discovered.
Competitive Landscape
This development positions Anthropic in closer rivalry with OpenClaw, an open-source platform that gained widespread attention recently. OpenClaw enables users to dispatch assignments through WhatsApp or Telegram for local device execution. Nvidia recently introduced NemoClaw, a corporate adaptation of the framework.
OpenAI entered this market segment last month by recruiting Peter Steinberger, OpenClaw’s developer, to develop personal AI assistant technology.
Currently, Claude’s computer control capability remains available as an experimental preview on MacOS exclusively for Pro and Max subscription tiers.
