- Gemini for macOS can now execute multi-step tasks across local files, desktop apps, and the web
- Users grant per-app access and can follow execution via a live preview window
- Rollout starts in English for US users, ahead of Apple's new full disk access restrictions
AI CURATED HOT STORY · 11/20/2026
Google Brings Computer Use to Gemini on Mac for Multi-Step Tasks
Google has launched computer use capabilities for its Gemini macOS app, enabling the AI to operate native Mac applications, browse the web, and execute multi-step tasks across local files in the background. Users can grant access to specific apps and preview the execution process in real time, maintaining control over the AI's actions. The feature is initially available in English to US users and requires a Google AI Plus, Pro, or Ultra subscription.
The rollout comes just before Apple imposes new restrictions on apps requesting full disk access, highlighting Google's strategic timing to navigate privacy compliance. While Google trails competitors like OpenAI, Anthropic, and Meta in launching computer use, integrating the capability directly into the Gemini desktop app offers a unified experience across native apps and web. This move intensifies competition in desktop AI agents and signals a shift from conversational to execution-oriented human-computer interaction. For developers, it introduces new automation workflows and security authorization paradigms; for users, it promises efficiency gains but raises privacy concerns that will shape industry discourse.
Hourly heat curve
2 fully observed hoursCoverage timeline
Total 2 reports · Latest first- The Verge AIT2·68 pts
- TechCrunch AIT2·62 pts
- Gemini's macOS desktop app gains computer use to operate native apps and websites in the background
- Initially U.S.-only and English-only, requiring a Google AI Plus, Pro, or Ultra subscription
- Google trails OpenAI, Anthropic, and Meta in shipping desktop agentic features