Google Messages is getting smarter scam protection with Gemini AI
Google is preparing to roll out a new scam-fighting feature for Google Messages called Enhanced protection. The optional setting is designed to help users spot suspicious texts from unknown senders, especially the more convincing scams that unfold slowly over time.
According to Google, Enhanced protection will begin rolling out in the coming weeks. The company has not yet confirmed which regions will receive it first, but the feature is part of a broader push to strengthen Android security against text scams, malicious links, call forwarding tricks, and screen-sharing fraud.
How Enhanced protection works in Google Messages
Google Messages already includes scam detection that can identify many suspicious texts directly on the device. Enhanced protection goes a step further by allowing conversations from unknown senders to be analyzed using Google’s cloud-based Gemini models.
The goal is to catch scams that are harder to detect with simple keyword or link scanning. Many modern text scams begin with casual, harmless-looking messages. The sender may build trust over several days or weeks before asking for money, personal details, verification codes, or account access. By reviewing the broader context of a conversation, Gemini may be able to recognize warning signs earlier.
Google says this cloud analysis happens through Private AI Compute, a system the company describes as a secure environment where data is encrypted and processed in a dedicated space. Google also says the content remains inaccessible to anyone else, including Google itself.
However, some privacy questions remain. Google has not clearly stated how long this data may be retained or whether independent third parties will audit the privacy and security claims. Because of that, Enhanced protection is being offered as an opt-in feature rather than something automatically enabled for everyone.
Gemini Nano will help check suspicious links on select phones
Google is also improving protection against dangerous links in messages. Scammers often create fresh websites that have not yet appeared on traditional blocklists, making them harder to catch with older security tools.
To address this, Gemini Nano will analyze unfamiliar links using real-time web signals directly on the device. This means some link checks can happen locally without sending the content to the cloud.
This advanced link-checking feature will be limited to newer premium devices, including the Pixel 10 and later Pixel models, the Galaxy S26 series, the Galaxy Z Fold8, the Galaxy Z Flip8, and other selected Android phones.
A redesigned warning system for links from unknown numbers will be available much more widely. Google says these warnings will work on devices running Android 6 and later. Previously, similar protections were restricted to certain countries or to users with Android Advanced Protection enabled.
What happens when you report spam in Google Messages
Google notes one important exception regarding message privacy. If a user manually reports a text as spam inside Google Messages, the message and its contents are sent to Google for review and analysis.
This reporting step is optional. It is available on the Pixel 6 and later, as well as the Galaxy S24 series and newer models.
Google says its scam detection features across Phone by Google and Google Messages already identify more than 5 million suspected scam calls and texts every month. That number comes from Google’s own reporting.
More Android protections are coming beyond Messages
Google’s anti-scam updates are not limited to texting. The company is also adding protections for screen sharing, call forwarding, and scam alerts in notifications.
One upcoming feature will hide login fields and payment screens in Chrome and Android WebView while a user is sharing their screen. This is meant to stop scammers from tricking people into revealing passwords, banking details, or payment information during a screen-sharing session. Google plans to roll this out in October for devices running Android 15 and later.
Another safeguard targets call forwarding scams. Fraudsters sometimes convince victims to enter special codes into the phone dialer, secretly redirecting incoming calls to another number. This can allow attackers to intercept one-time passcodes or account recovery calls.
Android will now block apps from enabling call forwarding in the background. The system will also show a warning before a call forwarding code takes effect if a user is persuaded to enter one manually. This protection is expected to arrive in early December on Pixel phones and other selected Android devices.
Scam alerts for more Android brands
Google has also been testing scam alerts inside notifications from popular messaging apps. Until now, this has largely been a Pixel-focused feature. Later this year, Google plans to expand these alerts to more Android brands, including Motorola, OnePlus, Oppo, Xiaomi, Vivo, Honor, and others.
This could make scam warnings more visible across the Android ecosystem, especially for users who rely on multiple chat apps rather than standard SMS messaging.
Should you turn on Enhanced protection?
Whether Enhanced protection is worth enabling depends on how you balance security and privacy.
If you often receive texts from unknown numbers, the new Google Messages feature could provide an important extra layer of protection. It may be especially useful against romance scams, fake delivery messages, investment fraud, impersonation attempts, and long-running social engineering attacks.
On the other hand, users who prefer to keep their message content away from cloud-based processing may want to leave Enhanced protection turned off. The existing on-device scam detection will still remain available.
Google’s new security tools show how AI is becoming a larger part of everyday smartphone protection. Gemini-powered scam detection could help Android users avoid increasingly sophisticated attacks, but its success will depend on user trust, transparency, and how well Google explains what happens to sensitive message data.
