Google DeepMind and the Government Technology Agency of Singapore (GovTech) announced an on-device scam detection model on October 14, 2026, built to warn smartphone users about manipulative conversations without sending private messages to the cloud, in a country where scams have accounted for more than S$4 billion in losses since 2019.
Scam Losses and the Shift to Social Engineering
According to the announcement on the Google DeepMind blog, authored by JJ Geewax and Miguel de Andrés, scams make up over half of Singapore’s reported crimes and have accounted for more than S$4 billion in losses since 2019. The announcement says scam tactics have moved from simple phishing to complex social engineering, in which attackers build convincing conversations to manipulate victims before asking for money, and that such scams now account for up to 80% of all losses.
Generic models often struggle with localised deception techniques, making scam detection harder, the announcement states. Google DeepMind’s lightweight Gemma model was fine-tuned on local scam conversations, ranging from impersonations of public institutions, banks, and local figures to trending regional schemes. The announcement says this enables the model to tell harmless social chatter apart from genuine manipulation, and that the resulting model is optimized to run on typical smartphones in Singapore.
On-Device Processing and a Two-Stage Pipeline
The model processes conversation data on the device itself. Conventional threat detection scans messages on servers in the cloud, the announcement explains, but many messaging platforms are end-to-end encrypted and therefore cannot inspect private conversations. It says running the analysis locally allows faster examination of conversation patterns to detect manipulative intent, so people can act before financial or credential exposure occurs.
When the model detects manipulative conversational patterns, a real-time warning appears on the device, urging the user to pause before sharing sensitive information or money.
Battery efficiency was a critical concern in running a complex scam classifier on modern smartphones, according to the announcement. The teams addressed this with a two-stage pipeline. In Stage 1, a lightweight, efficient classifier screens inbound messages using minimal power, focused on catching potential threats. In Stage 2, the fine-tuned Gemma model performs a contextual evaluation of a larger snapshot of the conversation to confirm a threat or discard a false alarm.
Governance Measures
The collaboration was presented at Singapore International Cyber Week 2026. The event’s opening address was delivered on October 13, 2026, by K Shanmugam, Singapore’s Senior Minister and Minister for Home Affairs, according to the address published by the Ministry of Home Affairs.
The announcement describes strict governance maintained throughout development under measures it labels Sovereign Sandbox, No Citizen Data, and Stress-Testing. Under Sovereign Sandbox, fine-tuning took place inside GovTech’s managed and maintained Google Cloud Platform environment, keeping training data isolated. Under No Citizen Data, training strictly excluded private or personally identifiable information and used anonymised datasets drawn from scam reports and intelligence feeds. Under Stress-Testing, the model was tested against synthetic dialogues and local corpora covering diverse scam categories and evolving impersonation tactics.
“Fighting scams shouldn’t mean sacrificing personal privacy,” said Mark Chen, Director of Trust and Safety at GovTech Singapore. “Working with Google DeepMind, we’re bringing advanced AI protection directly onto citizens’ devices. This allows us to catch highly targeted, local scam attempts in real time while keeping personal communications completely private.”
National AI Partnership Background
The scam detection model was delivered as part of Google’s National AI Partnership with the Singapore Government. Google DeepMind announced that partnership on May 20, 2026, launching new programmes in the country under its National Partnerships for AI initiative after opening a research lab in Singapore the previous year.
Those programmes include an exploration of AI co-clinician collaboration with public health clusters; support for infectious disease research and pandemic preparedness with Google.org, part of a $7 million contribution to the Philanthropy Asia Alliance’s Health for Human Potential coalition; a Gemma-powered running assistant for blind and low-vision athletes, tested and iterated with SG Enable, Singapore’s focal agency for disability and inclusion; training for local researchers on agentic AI science tools with the National Research Foundation; Gemini for Education provided to educators from primary schools to junior colleges; and research on multimodal and multilingual safety benchmarks with the Infocomm Media Development Authority and MLCommons.
Integration Path for Messaging and OS Providers
The announcement states that messaging platforms and operating system vendors can directly integrate these lightweight models to safeguard users locally, keeping chat interactions secure without exposing private data. It presents the project as a model for messaging platforms balancing user protection with end-to-end encryption, and as an example of governments and model providers working together to build technology for public good.
Credit: Source link
