Google Steps Up Fraud Fight With AI in Chrome and Android

Google has announced new AI-powered anti-fraud measures that will be rolled out across Chrome, Search, and Android. The company will begin using the Gemini Nano language model to improve Safe Browsing in Chrome 137 for desktop. The model runs on-device, allowing it to instantly identify fraudulent sites, even if you haven’t seen them before.

Gemini Nano analyzes the structure of web pages and quickly adapts to new deception schemes. The system is already being used to combat remote technical support fraud, where attackers trick users into giving up personal data by supposedly helping to fix problems. The AI ​​checks sites for signs of deception, such as the use of APIs to lock the screen, and passes the data to the safe browsing system.

Chrome developers emphasize that Gemini Nano works locally, without loading the device, thanks to optimized use of processor and memory resources. In the future, Google plans to expand the functionality to combat other types of fraud. Also this year, similar protection will appear in Chrome for Android.

At the same time, Google improved its system for detecting fraudulent sites using AI, which allowed it to block 20 times more malicious pages in search results. Thanks to this, the number of phishing attacks imitating government services decreased by 80%, and fake visa services by 70%. For Android users, Chrome has added another protection: AI on the device now warns about suspicious notifications that can lead to downloading malware or data leakage.

The system shows the sender, warns about possible fraud, and offers to unsubscribe. These updates follow the recent introduction of AI protection in the Messages app and a system for filtering fraudulent calls. In the future, Google plans to strengthen security in Android 16, adding features similar to Apple's protection: JavaScript blocking, 2G network disabling, device theft protection, and built-in antispam.