Singapore will get access to Google’s most advanced AI cyber defence tools, and GovTech Singapore is working with Google DeepMind on a scam detection model that runs on Android phones. Google announced both initiatives on 14 October, on the sidelines of Singapore International Cyber Week (SICW) 2026.
Fairwind gives defenders access to Gemini 4 Argon
Singapore is taking part in Google’s Fairwind Programme, which gives defenders access to the company’s most advanced cybersecurity tools. These include Gemini 4 Argon, Google’s new frontier model, and the CodeMender harness.
Google says the tools help defenders find, verify and fix vulnerabilities at scale. Fixing a flaw by hand can take weeks. With these tools, defenders can generate verified, deployment-ready patches in minutes, inside an organisation’s own secure cloud environment. The aim is to patch weaknesses early, before they disrupt essential systems and services.
On-device scam detection for Android
The second initiative pairs GovTech Singapore with Google DeepMind to build an on-device scam detection model for Android. It warns users about possible scam attempts in real time without sending their messages off the phone.
The reason is encryption. Scam detection has traditionally relied on scanning messages on cloud servers, but many messaging platforms use end-to-end encryption and cannot inspect private conversations. Moving the analysis onto the device gets around that. Google describes two parts to the approach:
- Local training: Lightweight Gemma models are fine-tuned on local scam narratives, including impersonations of public institutions, banks and local figures, as well as trending regional schemes. The model will keep being trained as scam tactics change.
- On-device only: The model analyses conversation patterns on the phone rather than on external servers. It shows a warning on the user’s screen, prompting them to pause before any financial loss or credential theft happens.
Google says the work shows local, real-time scam detection is possible without exposing private data. It has not said when users can install the model. More details are due “soon”.
Mark Chen, Director of Trust and Safety at GovTech Singapore, said: “Fighting scams shouldn’t mean sacrificing personal privacy.” He said the model lets the agency catch highly targeted local scams in real time while conversations stay private.
JJ Geewax, Director of Applied AI at Google DeepMind, said pairing DeepMind’s research with GovTech’s local domain and data expertise shows that digital safety and privacy can coexist on a device. He added that it could serve as a template for how governments and model providers work together.
Builds on earlier work with Singapore
The announcements follow the refreshed 2025 Memorandum of Understanding between Google and the Cyber Security Agency of Singapore (CSA), and the AI Agents Sandbox, which Google calls a global first, for studying how Computer Use agents behave in real-world settings.
Google also pointed to its existing consumer protections. It said enhanced fraud protection on Google Play Protect has blocked more than 3.3 million installation attempts of potentially malicious apps across more than 770,000 devices, and stopped 68,000 unique apps from being misused for financial fraud and scams.
On scam literacy, Google backs Be Scam Ready, an interactive game that teaches users to spot common manipulation tactics. According to Google, 10 minutes of play produced an estimated 20% lower overall risk of falling for a scam. Players were 47% less likely to mistake a scam for a legitimate message and 24% less likely to wrongly flag genuine communications. Google is also launching Be Scam Ready 2.0, which halves the playtime while giving the same level of protection.
