Adaptive Risk Analysis Framework for UPI Based Real-Time Payment Transactions

Digital payment platforms such as Unified Payments Interface (UPI) have witnessed exponential growth, accompanied by an increase in fraudulent transactions. Traditional rule-based fraud detection systems lack adaptability and fail to identify evolving fraud patterns in real time. This paper proposes SafePayAI, an adaptive fraud detection and prevention framework designed for real-time UPI transactions. The proposed system integrates Generative Adversarial Networks (GANs) for synthetic data generation and Random Forest classifiers for accurate fraud prediction. GAN-based data augmentation addresses class imbalance issues commonly present in financial datasets, while the Random Forest model enables robust classification using transaction behaviour and contextual features. The system assigns real-time risk scores to transactions and dynamically flags suspicious activities. Experimental results demonstrate improved detection accuracy, reduced false positives, and enhanced adaptability compared to conventional approaches, making SafePayAI suitable for real-world digital payment ecosystems.