Google Brings Gemini Auto Browse to Android Chrome

On August 18, 2026, Google rolled out Gemini in Chrome to all Android users in the United States. This release brings the built-in browsing assistant to mobile screens, allowing users to summarize articles, ask questions about web pages, and connect with Google Calendar and Keep. It also introduces image generation and editing on mobile devices through Nano Banana. Crucially, the update extends auto browse, an agentic tool that completes multi-step web tasks, to mobile devices for U.S. subscribers of the Google AI Pro and AI Ultra tiers.

Previously restricted to desktop Chrome, auto browse allows the browser to perform complex actions like booking event parking, updating recurring orders, or planning vacations. The agent operates by presenting a proposed plan for the user to review and approve before execution. To manage risks, the system pauses and requests explicit confirmation before submitting web forms, sending communications, scheduling events, or accessing financial and health data. The feature is limited to personal accounts for users aged 18 and older. Daily usage caps are tier-dependent: AI Pro subscribers are limited to 20 multi-step task requests per day, while AI Ultra subscribers can run up to 200.

For developers and security practitioners, the mobile expansion highlights Google's defense-in-depth approach to securing agentic browsers against indirect prompt injection. The security framework, first detailed in December 2025, utilizes a user alignment critic model to vet actions, origin restrictions, and a real-time prompt-injection classifier. Google has also established a bug bounty of up to $20,000 for demonstrated breaches of these agentic boundaries. However, Google warns that the feature remains experimental and can still make errors, such as unauthorized purchases, leaving ultimate responsibility with the user.

This mobile rollout follows the September 2025 desktop launch of Gemini in Chrome and comes shortly after the standalone Gemini app surpassed 1 billion monthly users in August 2026. While the mobile rollout is currently limited to U.S. Android users in English, it represents a major step toward embedding agentic AI directly into the default mobile browsing environment. The feature is not yet available on iOS or in live chats.