Anthropic Supercharges Claude AI With Computer Control Feature

Anthropic has announced a major upgrade to its Claude assistant, which allows it to directly control your computer to complete tasks. The feature is rolling out in research preview for Claude Pro and Max subscribers, hinting at its next approach, which appears to be toward more autonomous AI workflows. To catch you up, the announcement comes in the same month Anthropic upgraded Claude to create interactive charts, diagrams & visualizations in chat.

Claude moves beyond chat with computer control

With the latest update, Claude can point, click, scroll, and navigate apps on your system when it doesn’t have direct integrations available. It can open files, use a browser, and even run developer tools, all without requiring manual setup. If Claude cannot access a tool directly, it will use your screen just like a human would.

That being said, the system still prioritizes native integrations first, such as Slack or Google Calendar. Only when those are unavailable does it take over the screen. Every action requires user permission, and you can stop it at any point.

The feature also builds on Dispatch, which lets users assign tasks to Claude from their phone and continue them on a desktop. With computer control now added, Claude can actively complete those tasks in the background, whether that’s checking emails, generating reports, or handling development workflows.

Anthropic notes that this feature is still evolving, so it may struggle with complex tasks, and working through the screen is slower than direct integrations. The company has added safeguards like activity monitoring and prompt injection detection, while restricting access to sensitive apps by default.

For now, the feature is limited to macOS and requires the desktop app to stay active. However, support for more platforms is expected later, though timelines remain unclear. In related company news, Anthropic recently made a 1M token context standard for Claude 4.6.