CHENNAI: A New Frontier in Fraud Prevention

The National Payments Corporation of India (NPCI) is pioneering a 'federated model' in partnership with banks to tackle fraud within Unified Payments Interface (UPI) transactions. This initiative harnesses the power of Artificial Intelligence (AI) and Machine Learning (ML) to flag suspicious activities effectively.

According to Viswanath Krishnamurthy, NPCI's chief risk officer, this collaborative approach allows for the sharing of customer scores between NPCI and banks, combining transaction and device profiling with banks' demographic-based models. "This synergy is what we term as federated," Krishnamurthy explained.

Scaling New Heights in Transaction Security

Currently in pilot with 3-4 banks, the model promises to significantly reduce false positives, enabling more accurate fraud detection. With UPI processing 18.3 billion transactions worth Rs 24 lakh crore in March 2025, the stakes are high. Despite fraud being rare (1 in 6.5 lakh transactions), NPCI is leveraging AI/ML tools to further minimize risks, offering banks codes to identify and potentially block fraudulent transactions.

Investment scams account for 85% of UPI frauds, followed by fear-based scams (10%-12%), and those stemming from ignorance. Yet, UPI's fraud levels remain notably lower than credit cards, underscoring the platform's robustness and the potential of AI/ML enhancements.