Google Gemini Brings Background Computer Use to macOS Desktop

Google is bringing computer use to its Gemini app for macOS.

The assistant now operates across desktop applications and websites in a separate background window without taking over the user’s screen, The New Stack reported.

Gemini handles tasks in the background while users continue working. Users can watch the agent’s real-time progress through a chat preview, halt tasks when needed, and give approval for specific actions.

The announcement arrived on Thursday, just three days after Apple introduced tighter permission controls for macOS automation. The macOS 27.3 beta 2 updates divide Device Control and Data Access permissions into two distinct groups: System & App Protection Override and Accessibility.

Apps requesting broad system access now face more explicit permission prompts. Users can approve individual requests instead of granting ongoing access. Developers building desktop agents must also account for least-privilege access because users can restrict or revoke permissions while an agent runs. Google has not explained how Gemini will operate under these new restrictions.

According to Google, the Gemini model can pull shipping information out of local PDF files, pull together metrics for slide deck dashboards, process groups of images, and interact with 3D design applications. Computer use was first brought to the market by the company inside Gemini 2.5 back in October 2025 before being built straight into Gemini 3.5 Flash.

Developer documentation shows the model works from screenshots and returns actions for client-side execution. Developers manage the execution environment, while the API provides configurable safety policies and optional screenshot scanning for prompt injection. However, the new Mac feature is part of the consumer Gemini app rather than a separate developer API.

Google has not detailed how the background environment is isolated from macOS, how it maintains state across applications, or which specific actions require user approval. Questions persist regarding whether the app relies on identical execution architecture or if Google intends to grant third-party developers access to its background environment.

The company has documented prompt-injection safeguards for its developer-facing computer-use tools but has not disclosed which protections are built into the Mac app. Agents with access to local files and privileged applications risk untrusted input reaching the agent harness and influencing executed actions.

Reliability questions remain unanswered as the company has not shared performance figures for the Mac app. Google has also not explained what happens when an action fails or an application behaves unexpectedly.

Personal Google accounts owned by individuals aged 18 and older in the United States are currently receiving the initial rollout of Gemini computer use in English. To use the feature, individuals must have a Google AI Plus, Pro, or Ultra subscription alongside an Apple Silicon Mac equipped with macOS Sequoia 15.0 or a newer version.

The feature requires an Apple Silicon Mac running macOS Sequoia 15.0 or later.

Gemini computer use is rolling out to personal Google accounts belonging to users 18 and older in the United States, and requires a Google AI Plus, Pro, or Ultra subscription.

The assistant works across desktop applications and websites in a separate background window while users continue working, featuring a live preview in the chat for monitoring, pausing, and approving actions.