Google DeepMind Builds On-Device Scam Detector for Singapore

Google DeepMind and GovTech Singapore have jointly fine-tuned a privacy-preserving, on-device scam detection model using Google's lightweight Gemma. Scams account for over half of Singapore's reported crimes and more than S$4 billion in losses since 2019, with social engineering now driving up to 80% of all losses. The model runs entirely on-device, meaning neither Google nor the Singapore Government can access users' conversations. A two-stage pipeline screens messages efficiently before deeper analysis confirms threats. Training used only anonymised data within a sovereign sandbox, with no citizen data included. Phased deployment within applications is planned.