Anthropic Brings Background Computer Use to Claude Code and Cowork on Mac
Anthropic has enabled background computer use for Claude Code and Claude Cowork on macOS, available to Pro and Max subscribers. The feature lets Claude click, type, and open apps on a Mac without taking over the user's active cursor, following a similar launch by OpenAI's ChatGPT earlier in 2026.
Anthropic has upgraded Claude's computer use feature to run in the background on macOS, the company announced September 2, 2026. The capability is now live for Claude Code and Claude Cowork, available exclusively to Pro and Max subscribers.
Background computer use allows Claude to control a Mac—clicking, typing, and opening applications—without commandeering the user's active cursor. According to Anthropic, users can assign Claude a task on the desktop and continue working on something else while the agent operates independently in the background.
Computer use itself is not new for Claude; Anthropic has offered the capability on desktop previously. What's changed is that it no longer requires exclusive control of the user's mouse and keyboard to function, removing a major friction point for practical daily use.
Context: OpenAI got there first
OpenAI's Codex-turned-ChatGPT introduced background computer use on Mac earlier in 2026, according to 9to5Mac. Anthropic's rollout follows that precedent, bringing Claude's desktop agent capabilities in line with competing offerings from OpenAI.
The feature is currently restricted to macOS only, with no stated timeline for Windows or Linux support. Anthropic has not disclosed specific pricing changes tied to this feature—it is bundled into existing Pro and Max subscription tiers rather than sold separately.
Why background operation matters
AI-driven computer use tools have historically required users to hand over full control of their screen, making the agent's actions visually intrusive and effectively blocking any other work during a session. Running the agent in the background eliminates that tradeoff, allowing Claude to execute multi-step desktop tasks—file management, app navigation, form filling—while the user works in a separate window or application.
Anthropic has not published benchmark data or task-completion metrics for the background mode specifically, and details on reliability, error handling, or session length limits were not disclosed in the announcement.
What this means
This is a feature upgrade to an existing product, not a new model release—Claude's underlying model is unchanged. The significance lies in removing a usability barrier that made computer-use agents impractical for real multitasking. Anthropic is playing catch-up here: OpenAI already shipped background computer use on Mac earlier this year, and Anthropic's move signals that autonomous desktop control is becoming a baseline expectation for AI coding and productivity tools rather than a differentiating feature. Expect both companies to compete next on reliability and task scope rather than the basic ability to run in the background.
Related Articles
Anthropic Adds Explicit Song Lyric and Copyrighted Character Bans to Claude's System Prompt
Anthropic quietly added detailed new restrictions to Claude's published system prompts, explicitly barring song lyric reproduction and AI-generated images of copyrighted characters. The change follows closely on the heels of a lawsuit from Sony Music Publishing and Warner Chappell.
Anthropic Releases Claude Fable 5.1 and Mythos 5.1, Cuts Cache Pricing 75% But Output Tokens Jump 70%
Anthropic launched Claude Fable 5.1 and Claude Mythos 5.1, claiming the top spot on Artificial Analysis's Intelligence Index at 66. Cache-read pricing dropped 75% to $0.25 per million tokens, but a 1.7x increase in output token usage pushes net per-task cost up 20%.
Anthropic Releases Claude Fable 5.1, Claims 52.6% on New Terminal-Bench-Science Benchmark
Anthropic released Claude Fable (and Mythos) 5.1, claiming a 52.6% score on the new Terminal-Bench-Science 0.1 benchmark — up sharply from 24.7% for Fable 5. Independent testing shows the model's five reasoning levels produce dramatically different output token counts and costs for identical prompts, ranging from $0.10 to $3.30 per request.
Anthropic Launches API to Detect Watermarked Text From Claude Models
Anthropic is rolling out a watermark verification API that lets approved regulators, media outlets, fact-checkers, and enterprises check whether text was generated by Claude. The system builds on Google's SynthID text method and responds to EU AI Act watermarking requirements in effect since August 2, 2025.
Comments
Loading...