AI-Powered Protection: Google’s Gemini Nano Fights Phone Fraud

On May 14, 2024, Google held its annual Google I/O conference as scheduled. During the event, Google highlighted its latest and most advanced AI technologies and showcased their various applications. Interestingly, Google also introduced an AI-powered anti-fraud feature based on Gemini Nano. This feature alerts users to potential fraud risks during phone calls and encourages them to hang up.

Gemini Nano is a small-capacity language model developed by Google, notable for its ability to run on mobile devices. At the I/O conference, Google claimed that Gemini Nano can automatically recognize phone call content and match it against known fraud scripts. If Gemini Nano detects a high risk of fraud, it will issue an alert. During the presentation, Google demonstrated this feature: in the demo, a scammer posing as a bank employee called a user from an unknown number, claiming the user’s bank funds were at risk and should be transferred to a secure location. At this moment, Gemini Nano interrupted the conversation, correctly informing the user that the caller was likely a scammer and displaying a notification stating that banks never ask users to transfer funds elsewhere.

Furthermore, considering user privacy, Gemini Nano operates solely on the local device, ensuring a high degree of confidentiality. However, the exact release date of this feature remains uncertain, and Google indicated that the feature will be disabled by default.