Abstract
AI and Machine Learning in Fraud Detection play a critical role in securing digital payments and ensuring economic stability. As digital payment fraud escalates, costing billions globally, traditional models struggle to address increasingly sophisticated tactics such as phishing, account takeovers, and salami slicing. AI/ML-driven solutions, including graph-based anomaly detection, hybrid models (deep learning + knowledge-based systems), and ensemble methods, provide enhanced detection capabilities. These systems adapt to evolving threats, detect fraud patterns, and minimize false positives/negatives while maintaining transaction integrity. Emerging challenges include fraudsters exploiting AI agents, adversarial learning, and bottlenecks in digital systems. Metrics like detection accuracy, precision, and ROI validate the effectiveness of AI/ML systems in combating fraud. Ethical considerations and regulatory compliance remain crucial to standardize AI/ML deployment globally. Future research must focus on scalability, adaptability, and resilience to counter advanced fraud schemes.