Google DeepMind releases an on-device scam detection model built with GovTech
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
Google DeepMind releases an on-device scam detection model built with the Government Technology Agency of Singapore (GovTech) — a lightweight Gemma variant fine-tuned on local scam conversations, it looks for the manipulative conversation patterns used to impersonate public agencies, banks and local figures and for region-specific scam methods, and warns the user before sensitive information or money is handed over. It runs inside the device rather than sending anything to the cloud, and sees only the conversations the user has consented to
Why it matters
It came with background figures: scams account for more than half of reported crime in Singapore, losses since 2019 have passed S$4 billion, and conversational scams make up as much as 80% of the total loss. This is a case where on-device inference is used not as a performance boast but as the means of keeping a premise — that conversations are not sent to a server — and because the model was fine-tuned on local data together with the government, it also reads as a step toward sovereign AI
Source — Primary sources first; secondary reports are labelled “Press”
Officialdeepmind.googlehttps://deepmind.google/blog/on-device-ai-scam-detection-mddi/Added in Week 44, 2026 · See that week
On the map — ⑲ On-Device line
- Previous stop@huggingface/kernels
- This recordGoogle DeepMind
- Next stopRTX Spark laptops