TOOL UPDATES

Anthropic Launches Computer Use for Claude, Enabling AI to Navigate Web and Apps

M MegaOne AI Mar 24, 2026 2 min read
Engine Score 8/10 — Important

This update significantly enhances Anthropic's Claude with computer use capabilities, impacting a broad range of developers and enterprises. It's a crucial feature addition, making Claude more competitive and actionable for users.

Editorial illustration for: Anthropic Launches Computer Use for Claude, Enabling AI to Navigate Web and Apps

Anthropic has launched computer use capabilities for Claude, allowing the AI assistant to interact with web pages and desktop applications by clicking, scrolling, typing, and navigating — effectively giving Claude the ability to operate a computer on behalf of users. The feature, announced March 24, is available as a research preview for Claude Pro ($20/month) and Max subscribers.

Computer use transforms Claude from a text-based assistant into an agent that can perform multi-step tasks across applications. Users can instruct Claude to fill out forms, navigate websites, extract data from multiple pages, manage files, and complete workflows that would otherwise require manual mouse and keyboard interaction. The system works by taking screenshots of the screen, identifying interactive elements, and executing actions through simulated input.

A companion mobile tool called Dispatch, which debuted the week prior, pairs with computer use to enable remote task assignment. Users can send Claude tasks from their phone that the agent executes on their desktop — checking email, compiling reports, or managing applications while the user is away from their computer.

The launch positions Anthropic alongside OpenAI and Google in the emerging computer-use agent category. OpenAI demonstrated similar capabilities with GPT-5.4’s computer use features, and Google has integrated agent functionality into Gemini. The competitive dynamic centers on reliability: current computer-use agents work well for structured, repeatable tasks but struggle with unexpected dialog boxes, CAPTCHAs, authentication flows, and dynamic page layouts that vary between sessions.

Anthropic has designated this a research preview rather than a production feature, signaling that reliability and safety constraints remain active areas of development. The company noted that Claude will refuse to perform actions that could cause harm, access sensitive accounts without explicit permission, or bypass security measures. How effectively these guardrails work in practice — when an agent has the ability to click any button on screen — will determine whether computer use transitions from preview to default feature.

Share

Enjoyed this story?

Get articles like this delivered daily. The Engine Room — free AI intelligence newsletter.

Join 500+ AI professionals · No spam · Unsubscribe anytime

M
MegaOne AI Editorial Team

MegaOne AI monitors 200+ sources daily to identify and score the most important AI developments. Our editorial team reviews 200+ sources with rigorous oversight to deliver accurate, scored coverage of the AI industry. Every story is fact-checked, linked to primary sources, and rated using our six-factor Engine Score methodology.

About Us Editorial Policy