Government AI3 min
Google DeepMind and Singapore build an on-device Gemma model to catch scam messages
Google DeepMind and GovTech Singapore have fine-tuned a lightweight Gemma model on local scam conversations so that it can flag fraud on the phone itself, without messages leaving the device. A two-stage pipeline screens incoming texts with a small classifier before escalating suspicious threads to the fine-tuned model.
Google DeepMind (company blog)Read