Google is adding an optional layer of cloud-based AI analysis to Google Messages in an effort to detect scams that begin as ordinary conversations. The Verge reported on October 14 that the new “enhanced protection” setting will roll out in the coming weeks and use Gemini models to evaluate the apparent intentions of unrecognized senders.

The feature is aimed at slow-burn fraud, where a sender’s purpose may not become suspicious until a conversation is already underway. Google says the cloud analysis will extend its existing on-device detection so the system can recognize newer or more sophisticated attempts by scammers to conceal what they are doing.

A phone uses an on-device chip and an encrypted cloud chamber to inspect potential threats.
The system combines cloud-based conversation analysis with on-device checks for unfamiliar links.

The cloud component also creates an important privacy distinction from Google Messages’ current on-device approach. According to Google, user data will be encrypted during the analysis and processed inside a dedicated secure environment that neither Google nor anyone other than the user can access. Enhanced protection will be opt-in, rather than enabled automatically.

Google is pairing that cloud review with a second layer that stays on the phone. Gemini Nano will inspect real-time web signals associated with unfamiliar links before a user opens them, looking for indications that a destination is malicious. The Verge’s report does not specify which parts of a conversation are sent for cloud analysis, how long processing data is retained, or what independent verification supports Google’s access claims.

Three barriers protect a phone from call forwarding, screen exposure, and suspicious notifications.
Android’s wider security update also addresses call forwarding, screen sharing, and scam notifications.

Android’s broader update also targets scam techniques outside a message thread. The operating system will block apps from turning on call forwarding and notify the user when an app tries to do so. Chrome and Android WebView will automatically conceal sensitive information while a user is sharing a screen, a safeguard designed to reduce exposure during interactions in which someone else can see the display.

Google is expanding an existing chat-notification scam detector beyond Pixel phones as well. The Verge lists devices from Honor, Motorola, Oppo, OnePlus, Realme, Tecno, Vivo, Xiaomi, and other manufacturers as part of the planned expansion. Google says the feature will reach hundreds of millions of additional users later this year.

Taken together, the changes show Android moving from simple spam filtering toward protection that follows a suspected scam across messages, links, call settings, notifications, and screen sharing. The new system’s reach could make it a significant consumer-security deployment, but its effectiveness and privacy guarantees will depend on real-world performance once the opt-in cloud analysis begins rolling out.