On Thursday, Google unveiled a report detailing its ongoing efforts to combat online scams across its various platforms by leveraging artificial intelligence (AI). The report emphasizes the integration of AI technology in products such as Search, Google Chrome, and Android. The Mountain View-based company noted that advancements in generative AI are enabling the detection and prevention of new kinds of scams, with the on-device Gemini Nano AI model deployed in the Chrome browser to specifically address website-based scams.
Innovative Approaches by Google in the Fight Against Scams with AI
In a recent blog post, Google announced the release of its Fighting Scams in Search report. This comprehensive document outlines enhancements made to the company’s AI-driven scam detection systems, claiming the new technology can identify up to 20 times more potential scam websites than before.
Website-based scams typically employ fraudulent or compromised sites to trick users into divulging sensitive information, financial details, or login credentials. While the tactics employed by these scams may vary, they generally aim to deceive users through imitation, creating a false sense of urgency, or presenting offers that appear too attractive to ignore. Common examples include phishing sites, online shopping fraud, investment and cryptocurrency schemes, tech support scams, clone websites, and subscription traps.
Within the Search platform, Google has enhanced its classifiers for detecting and categorizing scams through AI improvements. With these advancements, the company is better positioned to de-rank scams before they reach users. A noteworthy example highlighted in the report is the increased incidence of scammers impersonating airline customer service representatives. Google reported that such scams have decreased by over 80 percent in Search due to these improvements.
In addition to its Search enhancements, Google Chrome has upgraded the Enhanced Protection mode of Safe Browsing. This mode now incorporates the Gemini Nano large language model (LLM) for desktop, augmenting user protection. The model identifies potential phishing sites and alerts Google to take appropriate action. The company indicated that Gemini Nano is capable of detecting even novel scam tactics by analyzing the complex structures of websites.
“Our intention is to extend this protective capability to Android devices, along with a broader range of scam types in the future,” stated Google.
Furthermore, improvements have been made to the threat protection systems on Android. Google Chrome for Android will introduce a feature aimed at safeguarding users from malicious website notifications. When the browser identifies a suspicious notification from a site potentially associated with scams, it will display a warning labeled “Potential scam detected,” along with options to unsubscribe or review the notification.
If users trust the website in question and suspect that the AI-driven feature has erred, they have the option to view the blocked content and can also choose to permit future notifications from that site.
