Skip to content
Theme
OnDevice Tools

AIAI Tools & Apps

Gemini computer use arrives on Mac for desktop automation

Google's Gemini app can now interact with files and applications on a Mac to complete multi-step tasks, rolling out first to US users in English.

Illustration of a Mac desktop with the Gemini app showing a computer use preview window

Key takeaways

  • Google's Gemini app for Mac now has computer use, letting the AI interact with files, applications, and web content to complete multi-step tasks.
  • Users can control which apps Gemini accesses and monitor its actions through a live preview window.
  • The feature is rolling out in English for US users, with no timeline for wider availability.
  • Apple is preparing new restrictions for apps with full disk access, partly in response to risks from AI agents with computer use capabilities.
In this article
  1. Gemini gains computer use on Mac
  2. How users control and monitor Gemini
  3. Security context and Apple's upcoming changes
  4. Why this matters for Mac users

Gemini gains computer use on Mac

Google has enabled Gemini computer use in its Mac app, according to The Verge. The AI assistant can now interact directly with files, desktop applications, and web content on a user's machine. The capability began rolling out on 19 November 2026. It lets Gemini complete multi-step tasks that previously required manual clicking and typing. The feature is initially available in English for users in the United States.

Google's Gemini app for macOS can now see and interact with the contents of a computer. The company describes the new capability as a way to handle multi-step tasks. These tasks span local files, desktop programs, and web content.

Users can describe what they want done. Then Gemini can execute the steps. This replaces manual clicking and typing with conversational commands.

The feature is designed for practical, everyday workflows. Google points to examples such as completing online forms. Other examples include building presentation slides and preparing product listings for sale. These are repetitive, multi-step activities. An AI assistant that can navigate a desktop environment could save noticeable time here.

Rather than copying and pasting information between applications, a user can ask Gemini to handle the process. The assistant can also fill in forms without manual data entry.

The computer use functionality represents a significant step forward. Previous versions of Gemini were limited to analyzing shared content. Now the assistant can actively open applications. It can read files from local storage. It can perform actions across the entire desktop environment. This marks a shift from passive analysis to active task completion.

How users control and monitor Gemini

Google has built in controls. Users decide which applications Gemini can access. This permission model means the AI does not have unrestricted access. Users can grant or restrict access to specific apps. This gives them confidence about what the assistant can do.

The company has also created a way to follow along as Gemini works. The app opens its own background window. This window displays a live preview of what the AI is doing. The preview appears alongside the chat conversation. A user can watch the AI navigate through applications and files in real time.

If something goes wrong, the user can see where the process broke down. Then they can intervene before it causes problems.

The feature is available now in English. It is available for users in the United States. No specific date has been announced for a wider rollout. No date has been given for additional language support either.

Security context and Apple's upcoming changes

Illustration: AI, Apple logo, security, computer and the labels AI Agent, Apple, Computer Use

The timing of Gemini's computer use launch is notable. Apple is preparing new restrictions for applications. These applications request full disk access on Mac. The company has signalled that AI agents with computer use capabilities pose potential security risks. Apple is adjusting the operating system's permission model. The goal is to give users more visibility and control.

A preview of the new settings appeared in the macOS 27.3 beta. This beta was released during the week of 19 November 2026. This suggests that Apple is anticipating a growing number of AI tools. These tools need deep system access. Apple wants to ensure users can make informed decisions. The decisions are about what those tools can do with their data.

Google's decision to launch computer use now is strategic. It positions Gemini as an early entrant in a growing category. The permission controls that Google is building into Gemini may serve as a model. The model is for how AI assistants can operate on desktop systems. The model also needs to maintain user security.

The live preview window offers a level of transparency. This transparency could help build trust. Some users may find AI agents unsettling. The preview shows exactly what the AI is doing at every step.

Why this matters for Mac users

For Mac users who already rely on Gemini, the computer use feature is useful. It extends the assistant's usefulness into the desktop environment. Tasks that previously required switching between multiple applications can now be handled. Tasks that required copying and pasting can also be handled. Tasks that required manual data entry can be handled too. All of this happens through a single conversational interface.

This could be particularly valuable for people. These people spend significant time on repetitive computer work.

The feature also highlights a broader shift. The shift is in how people interact with computers. As AI assistants gain the ability to operate software directly, the line between talking to a tool and using a computer begins to blur. Google's approach emphasises user control. It also emphasises transparency. The live preview window is part of this approach. This suggests one possible path for balancing convenience with security.

Until Apple's new restrictions are finalised, Gemini's computer use represents a first. It is one of the first mainstream attempts to bring agentic AI to the Mac desktop. How well it works in practice will likely influence other companies. Other companies may build similar features in the months ahead. The feature is available now for US users. These users need the latest version of the Gemini app on macOS.

Sources

  1. Gemini on Mac can now use your computerThe Verge

Tags

Spotted a mistake? See our corrections policy. Read how News by OnDevice works in our editorial policy.