Key Highlights

Anthropic introduced a capability allowing Claude to operate Mac computers from a distance to execute user tasks
Users delegate assignments via mobile; Claude launches applications, navigates online, and populates spreadsheets
This capability integrates with Dispatch, enabling users to delegate assignments to Claude through mobile devices
Anthropic cautions the technology remains experimental and recommends avoiding applications containing confidential information
This development follows OpenClaw’s viral success as an agentic AI platform, sparking competition across the technology sector

Anthropic has unveiled a groundbreaking capability enabling its Claude artificial intelligence to assume control of users’ computers for task completion. This enhancement was revealed on Monday and is currently accessible to Claude Pro and Claude Max members.

The functionality is exclusively compatible with MacOS. Users transmit instructions from their mobile devices, and Claude executes the assignment on their desktop machine — launching applications, operating a web browser, completing spreadsheets, or organizing files.

Anthropic demonstrated a scenario where a user behind schedule for a conference call requested Claude to convert a presentation deck into PDF format and append it to a calendar appointment. Claude accomplished the assignment without additional guidance.

The technology integrates with applications such as Google Calendar and Slack via native connectors. When a connector isn’t present, Claude can utilize keyboard and mouse controls manually to accomplish the assignment.

Claude consistently requests authorization before entering a new application. Users maintain the ability to halt any assignment at any point.

Integration With Dispatch Platform

The computer control capability is designed to function alongside Dispatch, a platform Anthropic introduced last week within Claude Cowork. Dispatch enables users to maintain ongoing communication with Claude across any device and delegate continuous assignments.

Combined, these capabilities can manage responsibilities such as distributing morning summaries, executing code evaluations, or monitoring emails automatically.

Anthropic acknowledges that sophisticated assignments may not succeed initially. The organization is positioning this as an experimental release to collect user input and determine priority areas for enhancement.

Privacy Concerns and User Advisories

Anthropic has implemented protective measures to minimize vulnerability, including automated detection for prompt injection threats. However, the organization continues to advise users against deploying the capability with applications managing confidential information — certain applications are deactivated by default.

Cybersecurity professionals have raised wider concerns regarding agentic artificial intelligence. These technologies can execute substantial actions rapidly and with minimal notification. Additional vulnerability exists for malicious actors to compromise the agent for accessing private documents or infrastructure.

Anthropic stated it will continuously enhance its protective measures as emerging vulnerabilities surface.

This enhancement positions Anthropic in stronger rivalry with OpenClaw, an open-source platform that achieved viral popularity earlier this year. OpenClaw enables users to transmit assignments via WhatsApp or Telegram and execute them locally on devices. Nvidia recently introduced NemoClaw, a commercial adaptation of the platform.

OpenAI similarly entered this domain last month, recruiting Peter Steinberger, the founder of OpenClaw, to develop personal AI assistants.

Currently, Claude’s computer control capability remains in experimental preview on MacOS for Pro and Max subscribers.

The post Claude AI Can Now Remotely Control Your Mac Computer From Your Mobile Device appeared first on Blockonomi.

---

Filed under: Bitcoin - @ March 24, 2026 11:16 am