Google Chrome Integrates On-Device AI for Enhanced Detection of Tech Support Scams

Google is introducing a significant security enhancement for Chrome, utilizing the ‘Gemini Nano’ large-language model (LLM) to better identify and mitigate tech support scams encountered during web browsing.

Tech support scams typically involve deceptive websites that mislead users into believing their devices are compromised by viruses or other issues. These scams manifest as full-screen alerts or persistent pop-ups that are challenging to dismiss, ultimately persuading victims to contact provided phone numbers for purported assistance. This often leads to unnecessary charges for remote support services or, in severe cases, unauthorized access to personal devices, resulting in potential financial loss or data breaches.

The new functionality in Chrome version 126 leverages AI capabilities integrated directly within the browser, facilitating rapid, privacy-conscious analysis of web content. This anti-scam mechanism, embedded within Chrome’s ‘Enhanced Protection’ feature, operates in real-time to assess web pages for known indicators of scams, including fraudulent virus warnings and full-screen lockouts.

Crucially, this analysis occurs offline on the user’s device using the Gemini Nano model. If a scam signal is detected, relevant data—including the outcome from the LLM and site metadata—will be communicated to ‘Google Safe Browsing’ for a comprehensive review. Should malicious intent be validated, Chrome will promptly alert the user about potential risks.

Google emphasizes that user privacy is paramount, stating that the entire process minimally impacts system performance. They have implemented controls to ensure the LLM’s activation is limited and executed locally, with resource management strategies including the optimization of token usage and asynchronous processing that prevents disruption to the browsing experience.

The rollout of this AI-driven security feature is set for Chrome version 137, which will be available to users upgrading to the latest version, provided they activate ‘Enhanced Protection’ in the Safe Browsing settings.

Users can enable this feature by navigating to Chrome Settings > Privacy and Security > Security > Enhanced Protection.

Looking ahead, Google has indicated intentions to broaden this security capability to identify additional fraudulent schemes, such as fake delivery notifications and toll scams, with the promise of extending the feature to Chrome for Android by 2025.

This initiative closely resembles a recent deployment by Microsoft for its Edge browser, which employs a specifically trained machine learning model to detect and block similar scams targeting users.