On Thursday, Google announced a new set of artificial intelligence-powered tools aimed at combating online scams across Chrome, Search, and Android.
The company said it will start using Gemini Nano, its on-device large language model, to enhance Safe Browsing in Chrome version 137 for desktops. According to Google, the on-device approach enables real-time analysis of suspicious websites, offering protection even against previously unknown scams. Gemini Nano is particularly suited for this task due to its ability to interpret the complex structure of websites and adapt swiftly to new scam tactics.
Google is already applying this AI model to detect remote tech support scams, which typically deceive users into revealing personal or financial data by presenting fake computer issues. The model scans web pages for indicators of scams, such as the use of the keyboard lock API. These signals are passed to the Safe Browsing system to assess the likelihood of a scam.
To optimize performance, the company said the model is triggered selectively and runs entirely on the device. Measures are in place to minimize resource use, such as limiting token usage, running processes asynchronously to avoid disrupting browser activity, and enforcing throttles to reduce GPU load.
Future updates will expand this technology to detect additional scam types, including those related to package delivery and unpaid tolls. These features will also be introduced to Chrome on Android later in the year.
As part of the broader announcement, Google shared that its AI systems now identify 20 times more deceptive pages than before. This has led to an over 80 percent reduction in scams impersonating airline support and a more than 70 percent drop in fake government or visa-related sites in 2024.
In addition, Google is launching a new notification warning feature in Chrome for Android. This system uses on-device machine learning to alert users when malicious websites try to send deceptive or spam-like notifications. When triggered, it displays the site name, a warning message, and options to unsubscribe or view the content.
These enhancements follow earlier deployments of scam detection tools in Android’s Messages app and similar features for detecting fraudulent calls introduced last year.
The update comes as Google prepares to introduce an Advanced Protection mode in Android 16. This feature will disable JavaScript, block 2G connections, and activate several security options by default, including Theft Detection Lock, Offline Device Lock, and spam filters in Messages.
Google is also reportedly developing a tool that detects scams during phone calls, particularly those that attempt to lure victims into accessing their banking apps, according to Android Authority.
Found this article interesting? Follow us on X(Twitter) ,Threads and FaceBook to read more exclusive content we post.
