Google DeepMind said it has fine-tuned a privacy-preserving, on-device scam detection model with the Government Technology Agency of Singapore (GovTech Singapore).[1] The model is built on Google's lightweight Gemma model, which was fine-tuned on local scam conversations, including impersonations of public institutions, banks and local figures.[1] Unlike server-side scanning of messages in the cloud, it processes conversation data directly on the device and triggers a real-time warning when it detects manipulative patterns.[1] Neither the Singapore Government nor Google will be able to access users' conversations through its deployment, the company said.[1] The rollout will be phased and developed with the Singapore Government.[1] Scams account for more than half of Singapore's reported crimes and more than S$4 billion in losses since 2019, the company wrote.[1]