Google DeepMind and Singapore are testing an AI tool that watches for scam conversations directly on your phone.
The tool comes from a partnership between Google DeepMind and GovTech Singapore, the country's government technology agency. It runs on lightweight, on-device Gemma models rather than cloud servers, and is built to flag conversational scams happening inside private chats. Because the analysis happens on the device itself, Google says no chat data leaves the phone to reach the cloud. The announcement describes this as a new tool being introduced through the partnership, not yet a nationwide rollout.
On-device processing is the real story here. Scanning private messages for scams usually means sending that data somewhere for analysis, which is exactly the privacy tradeoff users worry about. Running small models locally lets Google and Singapore make a scam-detection pitch without also asking people to trust a server with their texts.
Still, this is Singapore and Google testing a government-backed scam filter, not a feature rolling out to phones everywhere. Whether a model small enough to run on-device actually catches scams well in practice is the part worth watching next.