# Google DeepMind and Singapore put a fine-tuned Gemma model on phones to scan private chats for scams

A two-stage pipeline runs a cheap classifier on every inbound message and calls the Gemma model only on suspicious conversations, so no chat data leaves the device.

- Published: 2026-10-14T06:06:44.194Z
- Canonical: https://polylog.news/ai/2026-10-14/google-deepmind-and-singapore-put-a-fine-tuned-gemma-model-o
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Google DeepMind](https://deepmind.google/blog/on-device-ai-scam-detection-mddi/)

Google DeepMind and Singapore's Government Technology Agency (GovTech), working with the Ministry of Digital Development and Information (MDDI), have built an on-device model that detects conversational scams inside private chats with no cl…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-10-14/google-deepmind-and-singapore-put-a-fine-tuned-gemma-model-o (subscription information: https://polylog.news/pricing).