If your Android phone just offered to book your parking spot or reorder your dog’s food without you opening a single app, that’s Gemini Auto Browse, and it’s not a glitch. Google finished rolling it out to every eligible Android phone in the US on August 18, 2026, letting Chrome click through websites and fill out forms on your behalf while you watch. You still have to approve the plan before it acts, but the browser is doing the clicking now, not you.
- Full US rollout completed August 18, 2026, after a smaller test run that started at the end of June
- Requires a paid Google AI Pro or AI Ultra subscription — free Gemini accounts don’t get Auto Browse
- Needs Android 12 or newer with at least 4GB of RAM, US region, and English-US language set
- Handles multi-step chores: booking parking, updating recurring orders, organizing travel logistics
- Not available on iPhone, iPad, in Incognito mode, or on accounts under 18
What Gemini Auto Browse Actually Does
Think of it as handing Chrome a to-do item instead of a search query. Tell Gemini you need parking for a concert, and Auto Browse checks your ticket confirmation, pulls the venue and time, and searches a site like SpotHero for a spot near the address. It doesn’t just find options — it fills in the reservation form itself.
Google’s own announcement describes it working through errands like this: updating a recurring pet food order when your dog switches food, or pulling together flight and hotel options into one itinerary. A small floating icon marks the tab while Auto Browse is active, so you always know when the browser is driving instead of you.
This is a step past the Android feature drops Google has been shipping monthly — those add polish to existing apps, while Auto Browse changes what the browser itself is allowed to do on your behalf.
Who Can Use It Right Now (and Who Can’t)
Auto Browse isn’t a free-tier perk, and the two paid plans don’t get the same amount of it. Here’s how the limits actually break down before you go looking for the toggle.
| Tier | Auto Browse Daily Limit | Where It Works |
|---|---|---|
| Free Gemini in Chrome | Not included | Chat and page summaries only |
| Google AI Pro | Up to 20 multi-step tasks/day | Android only, US, age 18+ |
| Google AI Ultra | Up to 200 multi-step tasks/day | Android only, US, age 18+ |
Desktop Chrome has had basic Gemini chat since September 2025, and Google is widening that to more regions. Auto Browse itself, though, is Android-only for now, with no word yet on an iPhone or desktop version.
How to Turn On Auto Browse in Chrome for Android
If you already pay for AI Pro or Ultra, getting to Auto Browse takes about ten seconds. There’s no separate download or hidden settings menu to dig through.
- Update Chrome for Android from the Google Play Store to the latest version
- Tap the Gemini sparkle icon in the address bar, or open it from the three-dot menu
- Describe the task in plain language — for example, “find parking near the arena for my 7pm show”
- Review the plan Gemini proposes, then confirm before it starts clicking through pages
Because the rollout is staged, some AI Pro and Ultra subscribers may not see the option immediately even after updating. If it’s missing, checking again in a day or two typically does the trick.
The Safety Guardrails — and Where Google Admits They Can Fail
Letting an AI click through your accounts is the part that makes people nervous, and Google built in more than one layer of friction on purpose. A separate model, described as a “user alignment critic,” checks each proposed action before Auto Browse is allowed to run it.
- Confirmation required before purchases, form submissions, or visiting financial sites
- Built-in detection for prompt injection attempts hidden on malicious pages
- A visible on-screen marker whenever Auto Browse is actively controlling the tab
That said, Google isn’t pretending this is foolproof. The company’s own language calls Auto Browse “experimental” and warns it “can make mistakes or act unexpectedly, including completing a purchase without permission.” Worth reading twice before you hand it your Amazon login.
How This Fits Into Google’s Bigger Agentic Browsing Push
Auto Browse didn’t appear out of nowhere. Google has been layering agentic browsing into Chrome since I/O 2026, and it’s the same instinct behind the agent-style features OpenAI shipped in GPT-6 Astra earlier this year — every major AI lab is racing to make the assistant act, not just answer.
The difference is where the action happens. OpenAI and Anthropic mostly push agents through their own apps and APIs, while Google is building the agent straight into the browser most people already have open all day. That’s a much bigger surface area, for better and worse.
Frequently Asked Questions
What is Gemini Auto Browse?
It’s a feature inside Gemini in Chrome that completes multi-step web tasks for you, like booking parking or updating an order, by navigating and filling out pages on its own after you approve the plan.
Do I need to pay to use Auto Browse?
Yes. It’s exclusive to Google AI Pro and Google AI Ultra subscribers on Android in the US — there’s no free-tier access as of this rollout.
Is it safe to let Auto Browse handle logins and payments?
Google requires manual confirmation before sensitive actions like purchases, but the company itself warns the feature is experimental and can occasionally act unexpectedly, so treat it with the same caution as any new financial tool.
Will Auto Browse come to iPhone or desktop Chrome?
Not yet. As of the August 2026 rollout, it’s limited to Android phones in the US, though Google has been steadily expanding basic Gemini chat on desktop Chrome since late 2025.
Can Auto Browse buy something without asking me first?
It’s designed not to, and purchases are supposed to require your confirmation, but Google’s own documentation lists an accidental purchase as a known failure mode, so it’s worth double-checking receipts after you use it.
Should You Actually Turn This On?
If you’re already paying for AI Pro or Ultra, trying Auto Browse on something low-stakes — a parking spot, not a plane ticket — is a reasonable way to see if it earns a place in your routine. I wouldn’t hand it a payment method for anything that matters until Google drops the word “experimental” from its own description.
Leave a Reply