Introduction
Google has introduced an enhanced version of the Safe Browsing feature in Chrome 137, incorporating Gemini Nano [1], an on-device large language model (LLM) aimed at improving user protection against online scams [6]. This advancement significantly bolsters fraud protection across Chrome, Search [4], and Android [1] [4] [5], focusing on threats such as remote tech support scams, package tracking fraud [4], and unpaid tolls [4].
Description
Google has rolled out an enhanced version of the Safe Browsing feature in Chrome 137 by integrating Gemini Nano, its on-device large language model (LLM) designed to bolster user protection against online scams [6]. This AI-powered technology significantly enhances fraud protection across Chrome [4], Search [4], and Android [1] [4] [5], specifically targeting threats such as remote tech support scams, package tracking fraud [4], and unpaid tolls [4]. The integration strengthens the Enhanced Protection mode within the Safe Browsing service [2], which offers the highest level of security [5], reportedly providing twice the defense against phishing and other online threats compared to the Standard Protection mode.
Gemini Nano delivers immediate insights into risky websites by analyzing webpage content directly within the browser, marking a shift from server-side URL scanning to an on-device approach [5]. This capability allows for real-time detection and blocking of malicious campaigns [3], as it evaluates webpage intent by identifying scam-related triggers [3], including the misuse of keyboard-lock APIs commonly employed by scam sites. This on-device analysis not only ensures user privacy by preventing the transmission of sensitive data to external servers but also enables immediate threat detection as users navigate potentially dangerous pages. When Chrome suspects a page may be harmful [7], Gemini Nano facilitates instant analysis [1], blocking 20 times more deceptive pages and reducing impersonation scams by over 70% [4]. This proactive protection allows for the identification of previously unseen scams, including significant online threats [1].
In addition to website protection [1], new AI-powered warnings are being introduced for Chrome on Android to address the risks associated with malicious notifications from websites [1]. When a notification is flagged by Chrome’s on-device machine learning model [1], users receive a warning along with options to unsubscribe or view blocked content [1]. Users can also choose to allow notifications from sites they believe were incorrectly flagged [1], further enhancing their browsing experience and security. For those who opt into Enhanced Protection mode, summarized security signals indicating malicious intent are sent to Google [7], while users on Standard Protection will still benefit as flagged scam sites are added to the global Safe Browsing blocklist [7], aiding in the prevention of future scams [7]. Plans are underway to extend this AI-driven protection to additional scam types and explore further integration with Android, marking a significant advancement in real-time [3], adaptive threat detection while maintaining user privacy and performance [3].
Conclusion
The integration of Gemini Nano into Chrome’s Safe Browsing feature marks a significant advancement in online security, providing enhanced protection against a variety of scams. By shifting to an on-device analysis approach, Google ensures user privacy while enabling real-time threat detection. The introduction of AI-powered warnings for Android further strengthens this protection. As Google continues to expand this technology to cover more scam types and integrate further with Android, users can expect improved security and a safer browsing experience.
References
[1] https://blog.google/technology/safety-security/how-were-using-ai-to-combat-the-latest-scams/
[2] https://www.infosecurity-magazine.com/news/google-ai-gemini-nano-scams-chrome/
[3] https://gbhackers.com/chrome-137-integrates-gemini-nano-ai/
[4] https://thenimblenerd.com/article/google-unleashes-ai-avengers-scam-busting-breakthroughs-across-chrome-android/
[5] https://www.heise.de/en/news/Fraud-protection-Google-integrates-local-AI-model-Gemini-Nano-into-Chrome-10378672.html
[6] https://techcrunch.com/2025/05/08/google-rolls-out-ai-tools-to-protect-chrome-users-against-scams/
[7] https://securityonline.info/chrome-137-uses-on-device-gemini-nano-ai-to-combat-tech-support-scams/
