Anthropic's Claude Code now controls your Mac or Windows desktop. Here's what the computer-use feature does and the security risks.

DevOps publication provides insights into DevOps practices, tools, and culture, offering articles and tutorials for automating software delivery pipelines and fostering collaboration between development and operations teams. By exploring DevOps's curated content, developers can learn about infrastructure as code (IaC), containerization, and cloud-native technologies for building scalable and resilient applications. DevOps publication offers  guidance and best practices to accelerate your software delivery process and improve team collaboration.

DevOps.com

Anthropic has expanded Claude Code and Claude Cowork with a computer use capability that lets the AI directly control Mac and Windows desktops — clicking, typing, opening apps, and navigating browsers. The feature uses a layered approach: structured API connectors first, then browser control, then raw desktop control as a last resort. A companion feature called Dispatch lets users assign tasks remotely from a phone for Claude to execute on a desktop. Security concerns are significant, particularly around prompt injection attacks, and Anthropic itself warns the feature is early-stage and not yet enterprise-ready. Claude Code has surpassed $2.5B in annualized revenue, and the Windows launch came just ten days after the macOS debut.

Claude Code Can Now Run Your Desktop

Two Tools, Same Capability, Different Audiences