Skip to content
Trending storyDeveloping

Google DeepMind and Singapore GovTech launch on-device AI scam detection model

2 articles2 sourcessince Oct 13Last article 1h ago ·

Overview

AISummary of 2 articles

Google DeepMind and Singapore's Government Technology Agency (GovTech) have launched an on-device scam detection model, fine-tuned from Gemma on local scam conversations and optimized to run on typical smartphones in Singapore.

According to Google DeepMind, the model screens messages on the device, so end-to-end encrypted chats do not need to be scanned in the cloud, and it warns users before they share sensitive information or money.

The partners presented the work at Singapore International Cyber Week (SICW 2026) as a model for messaging platforms that need to protect users while keeping end-to-end encryption. Singapore's government reports that scams account for over half of the country's reported crimes and more than S$4 billion in losses since 2019.

Written by AI from the articles below · updated Oct 13, 10:35 PM ET

Check the sources:

Latest developmentOct 13, 10:05 PM ET
Google DeepMind builds on-device scam detection with Gemma and GovTech Singapore

Article timeline

The articles in this story. Times are ET.

Oct 13
  1. Google DeepMindOfficial
    Google DeepMind builds on-device scam detection with Gemma and GovTech Singapore

    AIGoogle DeepMind and GovTech Singapore are building an on-device fraud detection system based on the open Gemma model. The model is fine-tuned on local scam tactics to flag threats in real time without sending personal data off the device. Google DeepMind says over half of reported crimes in Singapore use sophisticated social engineering.

    GIF from @GoogleDeepMind's post
  2. Google DeepMindOfficial
    Google DeepMind and Singapore government launch on-device AI scam detection tool

    AIGoogle DeepMind and Singapore's GovTech built an on-device scam detection model, fine-tuned from Gemma on local scam conversations, to run on typical smartphones in Singapore. The model screens messages on the device, so end-to-end encrypted chats need not be scanned in the cloud, and it warns users before they share sensitive information or money. Scams account for over half of Singapore's reported crimes and more than S$4 billion in losses since 2019.

Heat trend

Not enough continuous observations to show a trend yet.