From AML Transaction Monitoring to Unified Financial Crime Analytics: Bridging Anti-Money Laundering and Bank Fraud Detection through Anomaly Detection, Graph Learning, and Risk-Based Alert Routing

Anti-Money Laundering (AML) transaction monitoring and bank fraud detection have historically evolved as distinct analytical and operational domains within financial institutions. AML systems are primarily designed to identify suspicious patterns indicative of money laundering or terrorist financing, emphasizing regulatory compliance, long-term behavioral deviations, and the generation of Suspicious Activity Reports (SARs). Conversely, fraud detection systems focus on short-term, real-time prevention of direct financial losses, leveraging supervised classification models and transaction-level risk scoring.Despite these operational differences, both domains rely on the same underlying transactional data, customer behavioral features, and risk indicators. Empirically, financial crime rarely occurs in isolation. Instead, it frequently manifests as a sequential criminal workflow: an initial fraud event (e.g., account takeover, social engineering scam, synthetic identity abuse) is followed by laundering strategies such as fund layering, structuring, mule-account distribution, or rapid cross-border transfers.