Google is finally rolling out the long-awaited AI-powered Scam Detection feature to its Pixel devices with the latest beta update for the Phone app. First showcased at I/O this May, the feature uses on-device AI models to identify and notify users of potential scam calls.

The Scam Detection feature can recognize conversation patterns typical of scam calls, like a caller claiming to be from your bank and asking you to transfer funds due to an alleged account breach. It then delivers an audio and haptic alert and a visual warning to prevent users from getting scammed.

To ensure privacy, Google says the feature processes the call entirely on the device and doesn’t send the conversation audio or transcripts to the company’s servers. In addition, the feature is off by default, and users have the option to enable it in the Phone app settings for a particular call or all calls.

During its initial announcement, Google claimed that the Scam Detection feature will be powered by the Gemini Nano on-device AI model. While this is true for the Pixel 9 series, the company has now confirmed that the feature will rely on other machine learning models on older Pixel devices. It’s not immediately clear if this will adversely affect the speed and accuracy of scam call detection.

Scam Detection is rolling out on the Pixel 6 and newer models as part of the latest Phone app beta release. The feature currently supports calls in English and is only available on Pixel devices in the US. At the moment, we don’t have any information regarding a wider release or additional language support.