During Google I/O, the Internet giant announced the upcoming deployment of a new anti-phone scam filter. Drawing on Gemini Nano AI, it will send an alert when the content of the call seems suspicious.

During Google IO the Internet giant announced the upcoming deployment

During Google I/O, the Internet giant announced the upcoming deployment of a new anti-phone scam filter. Drawing on Gemini Nano AI, it will send an alert when the content of the call seems suspicious.

A call from your bank asking you to validate a transaction? From your energy supplier who asks for your personal identifiers? Health Insurance asking you to give your credit card to a courier? This is probably yet another telephone scam intended to extract money from you. But these malicious calls are increasingly difficult to spot, with scammers constantly developing more and more sophisticated and personalized plans. And suffice to say that the consequences are proving costly… Over the last twelve months, the amount of losses suffered by individuals following scam telephone calls, fraudulent SMS messages, identity theft or investment fraud has reached nearly 1,000 billion euros, according to the GASA report (Global Anti-Scam Alliance).

In this fight against scams, AI can become a real ally, like Apate, the chatbot that scams scammers, or Scamio, a free chatbot capable of analyzing received communications and detecting attempts. of fraud. For its part, Google will integrate its Gemini Nano AI into certain smartphones, via Android 15, in order to detect potential telephone scams and alert users, as it announced during Google I/O. The AI ​​will display a real-time alert on the device in the event of a suspicious phone call. Initially, this option will be limited to certain Android smartphones, namely the Pixel 8 and the Samsung Galaxy S24, but it could be extended to other models later.

Google anti-scam filter: an alert in the event of a suspicious call

To offer this anti-scam filter, Google relies on Gemini Nano, the reduced version of its large language model which works directly in the smartphone, locally. The AI ​​will analyze the discussion in real time in order to detect the fraudulent language traditionally used by scammers, and therefore determine whether it is an attempted fraud or not. If the phone call is deemed suspicious, users will receive an alert during the call, prompting them to end the conversation.

During its presentation, Google used the example of the fake bank advisor scam, which asks its victim for personal information (passwords, bank card PIN codes, etc.), to make payments via gift cards or even validate money transfers (see our article). Gemini Nano then issued an alert on the Android smartphone.

The entire operation takes place locally, on the device, and not online, thanks to remote servers. This ensures that telephone calls are not stored elsewhere, and thus guarantees the protection of the user’s personal data as well as respect for their privacy. All conversations will be kept strictly private. Google should provide more details on this later. This anti-scam alert will be deployed during 2024, without further details.



ccn5