Google Launches Gemini Desktop Control for Mac, Sidestepping Apple's New Automation Guardrails
Google has rolled out computer-use capabilities to its Gemini app on macOS, allowing the AI assistant to operate across applications in the background. The timing raises questions about how the feature will comply with Apple's freshly announced tighter permission controls for system automation.

Google has introduced computer-use functionality to Gemini on macOS, enabling the AI assistant to execute tasks across desktop applications and websites within an isolated background window. The assistant operates without commandeering the user's screen, while a live preview in the chat interface lets users observe the agent's work, halt operations, and authorize specific actions as needed.
The announcement arrived Thursday, merely three days after Apple rolled out stricter permission controls for macOS automation. Google has yet to clarify how Gemini will function under these new restrictions.
Apple's tightened automation controls
macOS 27.3 beta 2 reorganizes Device Control and Data Access permissions into two distinct categories: System & App Protection Override and Accessibility. Applications seeking expansive system access will now encounter more granular permission prompts, permitting users to authorize individual requests rather than granting blanket, ongoing access.
Developers creating desktop agents must now design for least-privilege access, accounting for the possibility that users can restrict or revoke permissions while an agent is actively running.
Gemini's capabilities on macOS
According to Google, Gemini can pull shipping information from locally stored PDFs, compile metrics from dashboards for presentation materials, process batches of photographs for editing, and interface with 3D design applications.
Google has not disclosed technical details regarding the isolation of the background environment from macOS, the mechanisms for maintaining state across multiple applications, or which operations demand explicit user approval.
Google's computer-use foundation
Google introduced computer use with Gemini 2.5 in October 2025, subsequently embedding the capability directly into Gemini 3.5 Flash. The developer documentation indicates the model operates from screenshots and outputs actions for execution on the client side. Developers oversee the execution environment, while the API furnishes configurable safety policies and optional screenshot analysis to detect prompt injection.
The new Mac feature is part of the consumer Gemini app, not a separate developer API.
It remains unclear whether the consumer app leverages the same execution architecture or if Google intends to make its background environment available to third-party developers.
Unanswered security and reliability questions
Google has documented prompt-injection defenses for its developer-facing computer-use tools but has not specified which protections are embedded in the Mac app. Agents with access to local files and privileged applications face exposure to untrusted input that could reach the agent harness and alter its execution decisions.
Reliability questions remain unanswered since the company has not shared any performance figures for the Mac app or explained what happens when an action fails, or an application behaves unexpectedly.
Availability and requirements
Gemini computer use is rolling out in English within the United States for personal Google accounts of users aged 18 and older. The feature requires an Apple Silicon Mac with macOS Sequoia 15.0 or later and a Google AI Plus, Pro, or Ultra subscription.