Google Messages will use cloud Gemini to catch slow-burn scam chats

Google is adding an opt-in “Enhanced protection” mode to Messages that sends suspicious conversations to cloud-based Gemini models to catch scams that only reveal themselves several texts in. For Cybersecurity Awareness Month, Google announced the mode in a post on its blog, alongside several other Android protections. The setting rolls out in the coming weeks.

How the new mode works

The mode builds on Scam Detection, which Google launched in Messages as an on-device feature. Google says Messages will use cloud Gemini models to evaluate complex message context and more quickly understand a scammer’s intent in conversations with unknown senders. Engadget’s write-up gives Google’s example: a sender pretends to share a photo of the recipient with their dog, then keeps the conversation going and asks for the recipient’s name after being told it’s a wrong number.

Privacy is the obvious question, since this moves analysis off the phone. Google says Android analyses the messages in Private AI Compute, which encrypts them and processes them in a secure space that nobody else, Google included, can access. That’s a shift from the original design, where Scam Detection ran on-device by default. The new cloud layer is opt-in, and Google says it augments the existing on-device detection rather than replacing it.

Link safety gets its own upgrade. On-device Gemini Nano uses real-time web signals to judge whether an unfamiliar link is trustworthy before you tap. That check is available on the Pixel 10 and newer, the Galaxy S26 series, Galaxy Z Flip 8 and Z Fold 8, plus “other select devices”. The cloud analysis has a wider net: Droid Life reports it covers the Pixel 6 and newer, the Galaxy S24 and newer, the Galaxy Z Fold 6 and Flip 6 and newer, and other Android AICore devices. Messages also gets redesigned warnings that give extra time and clearer context before you tap links from numbers not saved in your contacts.

The rest of the package

Scam alerts in chat notifications, which were Pixel-only until now, are spreading to other brands. The warning appears inside the notification, so you don’t have to open the message. 9to5Google lists HONOR, Infinix, iQOO, itel, JOVI, Motorola, OPPO, OnePlus, realme, TECNO, vivo and Xiaomi, with those partners expected to enable it later this year. For readers in the Gulf, where those brands sell in volume, this is probably the part that reaches the most phones.

Two further changes target phone-based fraud:

• Screen-share privacy: the feature rolls out in October 2026 to devices running Android 15 or newer. By default, it hides passwords, banking details, and card security codes when you share a screen with Chrome or Android WebView open.
• Call-forwarding protection: it arrives in early December on Pixel and other select devices. It blocks apps from quietly enabling forwarding in the background, and Android warns you if a scammer talks you into typing an MMI forwarding code.

Google also described a bank-impersonation check in Brazil. Android verifies incoming calls with partner banks and hangs up if the caller fails, and Google says it has blocked over three million fake calls since June. It’s unclear whether the tool will come to other markets.

What it means in the UAE

Google hasn’t said which countries get Enhanced protection first, only that it’s coming “in the coming weeks”, so UAE availability is unconfirmed. Earlier this year, Google said it expanded Scam Detection in Messages to more than 20 countries, including Arabic.

That post doesn’t name the countries, though, and it covers the original feature rather than the new cloud mode. If you own a recent Pixel or Galaxy flagship, check Messages settings over the next few weeks. Google’s push to put Gemini across the OS is the same thread we covered when Gemini replaced Google Assistant on Android, and Google’s personal AI plans for Android point the same way. The original announcement is also summarised by Engadget and 9to5Google.