Anthropic Gives Claude AI Ability to Control Your Computer and Files

Anthropic announced that its Claude Code and Claude Cowork tools are being updated to accomplish tasks using your computer, capable of opening files, using the browser, and running developer tools. The move marks a decisive shift from AI as a conversational assistant to AI as an active operator of your machine.

When Claude lacks direct access to the tools it needs, it will point, click, and navigate what's on your screen to perform the task itself, opening files, using the browser, and running dev tools automatically with no setup required. When enabled, Claude will first prioritize connectors to supported services such as Google Workspace or Slack, but if a connector isn't available, it will still execute an assigned task.

The feature's origins stretch back to September 2024, when Boris Cherny, the Ukrainian-born creator of Claude Code and a former software engineer at Meta, asked his new tool a simple question: "What music am I listening to?" Claude opened his music player, snapped a screenshot, and returned the answer: "Husk" by Men I Trust. "I was taken aback," Cherny said. Claude Code is a command-line tool designed for developers, while Cowork brings the same agentic architecture to the desktop app, designed for non-coding knowledge work — no terminal required.

Claude Cowork, which was introduced in January, is designed for casual users and serves as an iteration of the Claude Code AI agent aimed at programmers. As Cherny described the design challenge: "We tried a bunch of different ideas to see what form factor would make sense for a less technical audience that doesn't want to use a terminal."

Claude computer use is initially available to Claude Pro and Claude Max subscribers on macOS. The feature is still in a research preview and will continue to be adjusted based on Anthropic's user feedback. The company also plans to bring the AI feature support to Windows x64 in future updates.

A companion feature called Dispatch extends the capability further. The capability pairs with Dispatch, released last week, which lets users assign Claude tasks from their iPhone and return to finished work on their desktop. Anthropic engineer Felix Rieseberg described the cross-device experience as "magical," citing example tasks such as generating reports from internal dashboards or finding a better seat on an upcoming flight. "Everything Claude can do on your computer — files, browser, tools — becomes reachable from wherever you are," Rieseberg said. Dispatch, a mobile tool that debuted last week, allows users to assign tasks to Claude from their smartphone, and have it complete those assignments using their computer.

Cowork conversation history is stored locally on your device, not on Anthropic's servers. Anthropic designed Dispatch to keep data local: processing happens on the desktop, the phone sends text instructions, the Mac does the work, and files stay on the user's machine. The desktop app must remain open and connected to the internet, and the computer must stay awake for remote tasks to run.

Guardrails are built into the system. Anthropic warned that "Claude can make mistakes," and said it has built the computer use capability with safeguards that minimize risk, with Claude always requesting permission before accessing new applications. Anthropic recommended not using this feature to handle sensitive information as a precaution. The company also implemented automatic scanning to detect prompt injection attempts.

"Computer use is still early compared to Claude's ability to code or interact with text," Anthropic acknowledged, noting the feature currently only works with Mac computers, meaning anyone with a Windows or Linux PC will have to sit this one out for now.

The launch comes as Anthropic sits at a $380 billion valuation after raising $30 billion from investors ahead of a possible IPO this year, with internal figures suggesting that between 70% and 90% of the code used to develop future Claude models is now written by Claude itself. Evan Hubinger, who leads Anthropic's alignment stress-testing team, put it plainly: "Recursive self-improvement, in the broadest sense, is not a future phenomenon. It is a present phenomenon.