Google & Singapore launch anti-scam AI initiatives

Google has announced two new cybersecurity and anti-scam initiatives with the Singapore Government, extending an existing collaboration on digital security in Singapore.

The first brings Singapore into Google's Fairwind Programme, giving cyber defenders access to tools including the Gemini 4 Argon model and the CodeMender harness. The second pairs GovTech Singapore with Google DeepMind to develop an on-device scam-detection AI model for Android devices.

The Fairwind Programme is designed to help security teams identify, verify and fix software vulnerabilities more quickly. Google says the system can generate verified patches in minutes within an organisation's secure cloud environment, instead of relying on manual processes that can take weeks.

The aim is to reduce the window in which software or system weaknesses can be exploited, helping protect essential systems and services from disruption and cyber attacks.

Scam detection

The anti-scam initiative focuses on analysing messages directly on a user's device rather than in the cloud. This is intended to address a practical challenge for threat detection on modern messaging services, many of which use end-to-end encryption and therefore do not inspect private conversations on central servers.

Google DeepMind and GovTech say the Android model is being built to recognise conversational patterns linked to scam attempts and display real-time warnings on screen. The model is based on lightweight Gemma models fine-tuned on local scam narratives, including impersonation of public institutions, banks and local figures, as well as regional scam trends.

According to the details released, the model will also be updated to reflect changing scam methods. The goal is to distinguish ordinary exchanges from the manipulative, urgent language often associated with social engineering fraud.

By keeping analysis on the handset, the system is intended to preserve the privacy of chat content while still flagging suspicious interactions. The project has so far been described as a proof of local, real-time scam detection, with further details on installation and use to come later.

Singapore has often served as a test bed for digital trust and cyber policy initiatives, and these measures add to a broader pattern of cooperation between Google and public agencies in the city-state. The announcement follows a refreshed 2025 Memorandum of Understanding with the Cyber Security Agency of Singapore, or CSA, on national cyber resilience.

It also builds on work in an AI Agents Sandbox set up to examine how computer-use agents operate in practical settings. That programme is part of a broader effort to understand the risks and controls needed as more AI systems are deployed in live environments.

Broader context

The initiatives come as governments and technology companies face growing pressure to tackle fraud, malware and increasingly automated cyber threats. Scam prevention has become a central policy issue in Singapore and elsewhere in Asia, where messaging apps, impersonation tactics and cross-border criminal networks have contributed to persistent consumer losses.

Google also pointed to existing consumer safeguards in Singapore through Google Play Protect. It says enhanced fraud protections have blocked more than 3.3 million installation attempts of potentially malicious applications across more than 770,000 devices, and prevented 68,000 unique apps from being misused for financial fraud and scams.

Alongside technical tools, Google says it has supported public education through Be Scam Ready, an interactive game designed to teach users how to identify manipulation tactics. Studies cited by the company found that 10 minutes of gameplay reduced the estimated overall risk of falling victim to a scam by 20% and made players 47% less likely to mistake a scam for a legitimate message. Players were also 24% less likely to incorrectly flag genuine communications.

A shorter version of the game, Be Scam Ready 2.0, is also being launched with the same stated level of protection in half the play time.

Mark Chen, Director, Trust and Safety, GovTech Singapore, described the device-based approach as a way to balance fraud prevention with privacy.

"Fighting scams shouldn't mean sacrificing personal privacy. 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," Chen said.

JJ Geewax, Director, Applied AI, Google DeepMind, said the work reflects the need for more sophisticated responses as fraud methods evolve.

"As scam tactics evolve, protecting people requires new, more advanced, more intelligent approaches. Pairing Google DeepMind's research with GovTech Singapore's local domain and data expertise means we can confront this urgent public safety challenge head-on. This initiative proves that digital safety and personal privacy can coexist on-device, offering a blueprint for how governments and model providers can collaborate to build technology to keep people safe," Geewax said.