Overview
- Google announced on Thursday, Nov. 19, 2026, that Gemini’s new “computer use” can click, type, navigate menus, and fill forms across local files, native Mac apps, and websites in an isolated background window.
- The feature began a U.S. English rollout for Google AI Plus, Pro, and Ultra subscribers and requires an Apple Silicon Mac running macOS Sequoia 15.0 or later and a personal Google account for users 18 or older.
- Users grant app-by-app permissions and see a live preview in chat so they can watch, pause, stop, or approve key actions, and Google says the agent will pause for logins and payments to require extra confirmation.
- Security and privacy warnings in Google’s docs and reporting note that enabling macOS Accessibility and Screen & System Audio Recording lets Gemini see active sessions and on-screen content, which can expose visible MFA codes and open the door to prompt-injection or cross-app manipulation.
- The launch follows similar agent tools from OpenAI, Anthropic, and Meta, and it arrives as Apple is adding tighter macOS permission controls, which could affect how the industry limits persistent or broad agent access to users’ machines.