DeepMind and Singapore Team Up on On-Device Scam Detection Model

DeepMind and Singapore's GovTech are working together on a specialized model for spotting scam conversations. The design relies on compact Gemma models running on the device, so private chat data does not need to leave the phone for cloud processing.
The collaboration brings DeepMind together with Singapore’s GovTech to develop a specialized model for recognizing scam conversations. The work centers on compact Gemma models designed to run directly on a device.
Because the model operates on the device, private chat data does not need to be sent to the cloud for processing. That design keeps the analysis local while targeting a specific use case: spotting scam exchanges in conversations.
This work could affect people who rely on chat apps, especially those targeted by scams, by enabling warnings without sending message content off-device. It may also shape expectations around privacy and on-device AI, though its real impact will depend on accuracy, adoption, and safeguards.