Anthropic’s Claude AI Gains Ability to Control Your Mac and Complete Tasks Remotely

Anthropic has expanded the capabilities of its Claude AI offerings by introducing a new 'computer use' feature that allows Claude Code and Claude Cowork to directly control a user's Mac desktop. This includes interacting with the mouse, keyboard, and screen to complete requested tasks like file management, browsing, and executing development environments. When possible, Claude prioritizes using direct app connectors such as Slack or Google Calendar, but falls back to desktop control if no connector is available, always requesting explicit user permission before performing actions. The new feature integrates with a tool called Dispatch, allowing users to send commands remotely from a phone to their Mac as long as the machine remains powered and awake. While promising, Anthropic cautions users about the inherent imperfection and potential errors during tasks, advising to start with trusted applications and avoid sensitive data during this research preview. Safeguards have been put in place to limit access to high-risk apps and to prevent risky operations like financial transactions and sensitive data collection, although the company acknowledges that these protections are not absolute. This release joins a growing wave of AI agents by companies like Perplexity, Manus, Nvidia, and the popular OpenClaw tool, highlighting a competitive race to deliver comprehensive, assistive AI desktop agents.