ToolsNews
Gemini on Mac gets computer use in a background window
Since November 19, the Gemini app for Mac can click, type and fill in forms across local apps and the web. It works in a separate background window, so your own screen stays free. For now it is US-only, in English, for subscribers.
· 3 min read

What launched?
On November 19, Google introduced computer use in the Gemini app for Mac (Google). You describe a task, and Gemini navigates menus, types, clicks and fills in forms across your local files, desktop apps and the web.
The key detail is that it does not take over your screen. It works in an isolated background window (a virtual screen) with a live preview in the chat, while you keep using your own mouse and keyboard.
How it works
- In Task mode, you start it by mentioning @Computer use in the prompt.
- You choose which apps it can access, you can stop it with one click, and it asks for approval before key actions.
- It pauses for logins and payments and may ask for confirmation before sensitive actions.
- It needs the Accessibility and Screen & System Audio Recording permissions. With on-demand file backups turned on, it backs up files before changing them (
- [object Object]
- ).
- According to 9to5Google, new desktop connectors such as Blender and Autodesk Fusion join the existing Git and Obsidian integrations (
- [object Object]
- ).
Google's examples: batch color-grading photos in design software, pulling weights and addresses from local PDFs into a shipping quote form, building a Google Slides deck from live dashboards, and staging resale listings from clothing photos.
Who can use it?
For now it is rolling out in English in the US to Google AI Plus, Pro and Ultra subscribers aged 18+ with personal Google accounts. It requires an Apple Silicon Mac running macOS Sequoia 15.0 or later. Work and school accounts are not supported.
Google's own risk list
The help page is unusually candid about the risks (Gemini Apps Help):
- Gemini has access to logged-in accounts and active sessions in native apps, not just the browser.
- Untrusted content in one app can manipulate it into jumping to another allowed app and acting there. That is classic prompt injection, now across the whole desktop.
- Allowing code editors or developer tools can enable shell execution and broader file system changes.
- It can read MFA prompts or codes visible on screen, for example in a messaging app.
- Some actions are irreversible because of the app, and backups do not always help.
Why it matters
Desktop computer use has mostly been a developer and enterprise agent feature so far (OpenAI brings computer-use agents to healthcare teams). The background window solves a practical problem: you do not have to watch the agent take your mouse away. The risk list, though, makes clear that such an agent has roughly the same permissions as you. If you try it, only allow the apps the task really needs, and keep it away from the terminal and from the app where your MFA codes arrive.
Produced with the help of AI, reviewed by an editor. How we make our articles
Reviewed and edited by the Neural Daily desk. About


