Google is giving Gemini access to your inbox if you let it. An opt-in toggle called Enhanced protection rolls out to the Google Messages app over the coming weeks, designed to catch slow-burn conversational scams that slip past standard on-device filters.
Most junk texts are crude enough for local filters to flag right away. Conversational fraud works differently. Attackers pretend to text the wrong number, strike up polite chatter, and only ask for money or credentials days later. To spot that buildup, Google Messages will send threads from unknown senders to its remote Gemini systems to evaluate the broader conversation.
Google says this cloud analysis happens inside its Private AI Compute architecture, where customer data is isolated in secure hardware enclaves and kept out of reach of anyone, including Google staff. What the company has not shared is how long those messages remain on its servers, or whether an outside security firm has verified those privacy claims.
Leaving the toggle off keeps your chats on your handset, relying solely on local heuristics. But if you turn it on, Google pairs the cloud inspection with fresh on-device safety tools.
Hardware requirements and wider Android defenses
Google is also using Gemini Nano to police suspicious web links. Rather than relying purely on static blocklists, Nano examines links from strangers against live web indicators directly on the handset. That feature has strict hardware limits. It only works on the Pixel 10 family, Samsung's Galaxy S26 lineup, the Galaxy Z Fold 8, the Galaxy Z Flip 8, and a few unspecified handsets. Older phones are not left out entirely: updated link cautions for texts from unsaved contacts are heading to phones running Android 6 and newer.
According to Google, existing scam tools across Messages and the Phone by Google dialer already flag more than 5 million suspicious calls and messages each month. That is an internal company metric that has not been independently audited.
Several companion defenses are arriving alongside the Messages changes. Starting in October, Chrome and Android WebView on Android 15 and up will automatically mask login boxes and credit card fields whenever you share your screen. In early December, Pixel phones and select rivals will block background apps from activating call forwarding, while warning users if an incoming caller tries to trick them into dialing an MMI forwarding sequence.
Scam warnings inside chat app notifications, previously exclusive to Pixel owners, will also roll out to hardware from OnePlus, Motorola, Xiaomi, Oppo, Vivo, and Honor before the year ends.
SDFiled by The Software Desk
Apps, platforms, operating system releases, security fixes and the features arriving in them.
More from this desk →
Be the first to comment
Join the argument. No password, just your email or a passkey.