Google is pushing Gemini deeper into Android with a new set of AI-powered features aimed at everyday phone tasks. The biggest change is a beta automation feature that can handle some multi-step actions in supported apps, such as ordering a ride or food delivery.
The rollout is not a broad release for every Android phone or every app. It begins with a limited set of categories, countries, apps, and devices, while Google adds safeguards meant to keep the user in control as Gemini acts on the phone.
Gemini starts taking on multi-step Android tasks
The new automations are designed to let users hand off parts of a to-do list to Gemini. Instead of only answering questions or opening an app, Gemini can move through some tasks that require more than one step.
Google says the beta feature will initially work with select apps in the food, grocery, and rideshare categories. That means the first version is focused on practical consumer tasks rather than a wide-open system that can operate across the whole phone.
In the U.S. and Korea, supported apps include DoorDash, Grubhub, Instacart, Lyft, McDonald’s, Starbucks, Uber, and Uber Eats. In Korea, Kaemin and Kakao T will also be supported.
At launch, the feature is limited to the Gemini app on certain devices. Those devices include the Pixel 10, Pixel 10 Pro, and Samsung Galaxy S26 series.
The launch is deliberately narrow
The limits matter because AI automations can create real friction if they misunderstand a user or take the wrong path inside an app. Google is not presenting this as a universal Android agent. The first version is a controlled beta tied to specific apps and specific phones.
The geographic rollout is also restricted. The feature will initially be available only in the U.S. and Korea, even though Android is a global operating system and Gemini already appears across parts of Google’s mobile ecosystem.
For users, the practical takeaway is simple: Gemini automations are arriving, but most Android owners should not expect immediate access. Availability depends on location, device, and whether the app category is part of the first supported group.
Google adds guardrails for AI actions
Because the feature can perform multi-step tasks, Google has added protections around how automations begin and how they run. An automation cannot start without an explicit command from the device’s owner.
Google also says users can watch the task progress in real time. If Gemini appears to be making an error or becomes stuck, the user can stop the task while it is running.
The company is also separating the automation from the rest of the phone. The task takes place in a secure, virtual window on the device, where Gemini can access only limited apps rather than the rest of the data on the device.
Those constraints help define what this feature is and is not. It is not a general promise that Gemini can freely control Android. It is a limited automation layer for approved use cases, with user confirmation and live oversight built into the flow.
Android gains more scam detection features
Google is also expanding scam detection across calls and texts. A Scam Detection feature for phone calls is becoming available on Samsung Galaxy S26 series devices in the U.S.
The phone-call feature is already offered on Pixel phones in the U.S., Australia, Canada, India, Ireland, and the U.K. That makes the Samsung Galaxy S26 series expansion another step in bringing the protection beyond Google’s own Pixel devices.
Google is also using its Gemini on-device model to detect scam texts in the U.S., Canada, and the U.K. on Pixel 10 series devices. The same scam-text detection is coming soon to Galaxy S26 series phones as well.
The important distinction is that Google describes the text detection as using a Gemini on-device model. The source does not provide more technical detail, but the placement of this feature alongside call protection shows that scam detection is becoming a central part of the Android AI update cycle.
Circle to Search gets broader screen awareness
The third major update affects Circle to Search, the Android feature that lets users start searches with gestures such as scribbling or circling. Google says Circle to Search can now search for everything visible on the phone screen, rather than only a single object.
That changes the feature from a one-item lookup into a broader screen-level search tool. Google gives the example of identifying every item of clothing and every accessory in an outfit someone likes, or learning more about a group of things and the related topic visible on the screen.
The update fits the same overall direction as the Gemini automations: Android is becoming more context-aware, and Google is trying to make AI features operate closer to what users are already doing on their phones.
Google has been releasing Gemini updates into Android through operating system updates and updates aimed at its flagship phone, the Google Pixel, including frequent updates known as Pixel Drops. The latest release continues that pattern, while Apple has been struggling to release a more comprehensive AI feature set that includes an AI-powered Siri, a launch that was recently pushed back again to later in the year.