Gemini in Chrome has finished rolling out to Android users in the United States, three months after Google first previewed it. If you are in the US, the spark icon should now be sitting in Chrome on your phone without you doing anything — and for Google AI Pro and Ultra subscribers, so is auto browse, the agentic mode that carries out multi-step tasks on websites for you.
Where to find Gemini in Chrome on Android
There are two entry points, per 9to5Google’s walkthrough of the finished rollout. The first is a toolbar shortcut to the right of the address bar — you can only have one shortcut there, so Gemini competes with whatever you already keep in that slot. The second is the three-dot overflow menu near the bottom of the screen.
Tapping the spark icon slides up a sheet that closely mirrors the desktop version of Gemini in Chrome. You can slide the sheet down and leave it docked at the bottom of the screen rather than dismissing it outright. From there it will summarise the page you are on and answer questions about what you are looking at, and Nano Banana integration handles image editing inside the same sheet. A tools menu turns Personal Intelligence on or off — that is the switch that decides whether Gemini can reach into Calendar, Keep and other connected Google apps.
Your current page can go into the prompt as context
This is the part worth pausing on. Per 9to5Google’s description, the page you are currently viewing can be added to your prompt as context, and there is an option in Settings to disable that behaviour. The existence of a disable switch is a strong hint about which way the default runs, though Google has not published a help page spelling out the data flow that we could find — so treat it as a setting to check rather than a settled fact.
Most of the time that is exactly what you want; it is the entire point of asking a browser assistant about a page. But “whatever tab is open” also covers a patient portal, a bank statement, a payslip or a half-written email in a webmail tab. Anyone who browses anything sensitive on their phone should find that Settings toggle first and decide deliberately, rather than finding out later what was already being sent. It is a two-minute check that costs you nothing if you decide to leave it on.
Auto browse is subscription-only
The genuinely new capability here is gated. Auto browse — where Gemini works through a task on a site instead of just talking about it — is limited to Google AI Pro and Ultra subscribers. Google’s examples are booking parking for an event, updating a recurring online order and organising travel, described as a way to “handle tedious everyday tasks for you”. Google’s stated guardrail is that sensitive actions such as purchases require explicit confirmation from you before they go through.
That confirmation step matters more than the marketing does, because an agent with your logged-in sessions is an agent that can spend your money and change your bookings. Treat the first few tasks as supervised, not delegated. We looked at how the same capability behaves on desktop, including what it does with your saved Chrome passwords, in Gemini Spark Can Now Use Your Chrome Logins and Saved Passwords.
The rollout is US-only for now, and no other markets have been named as part of it.
Sources: 9to5Google
