Google DeepMind and Singapore’s GovTech announce a new AI tool aimed at protecting citizens from scams on Android. The initiative focuses on detecting “conversational scams” occurring in private chats, using AI designed to run directly on users’ devices.
According to the DeepMind and Straits Times reports, the system uses lightweight on-device Gemma models to identify scam indicators without sending private chat content to the cloud. The tool is described as intercepting scam attempts while keeping data local, with “zero cloud data egress” intended to reduce exposure of sensitive communications.
The outlets emphasize different aspects: the Straits Times frames the project as a first-of-its-kind scam detection model for Android users, while the DeepMind announcement highlights the privacy and deployment approach—on-device processing, conversational scam targeting, and minimizing reliance on cloud infrastructure.
