Google DeepMind and the Government Technology Agency of Singapore (GovTech Singapore) are working together on a locally trained AI model that detects scams directly on people's phones.

Built on Google's lightweight Gemma model and fine-tuned on local scam conversations, it works in two stages: a simple classifier first screens incoming messages, then the model examines a larger part of the conversation to confirm a threat and warn users before they share money or personal details. Because messages are analyzed on-device without being sent to the cloud, neither the Singapore Government nor Google can access them, and the model will be rolled out in phases.