Ask Sumsubers: How has transaction monitoring evolved in recent years, and what emerging trends are shaping its future?

If you want to see how transaction monitoring has evolved over the last decades—look at financial crime and how it has changed. It’s no novelty to say that criminals are among the smartest people in the world. Oftentimes, regulators and monitoring systems have rushed to investigate what has already happened, adjusting their systems reactively. The trend now is to change that and make predictive modeling stronger. Let’s take a look at how transaction monitoring has developed.

It’s important not only to focus on how tech has evolved—but also how attitudes and approaches have changed over the years. The beginnings of transaction monitoring can be traced back to the Bank Secrecy Act (BSA) of 1970—a landmark US legislation which mandated reporting and other requirements on financial businesses to help detect and prevent money laundering (ML). This was a time when transaction monitoring was much less sophisticated than it is today. The early approaches relied on fixed thresholds (usually reporting of transactions exceeding $10K) and basic pattern recognition to flag suspicious activity—which often overwhelmed compliance teams with false positives and could miss increasingly sophisticated ML tactics. Seems simplistic as we navigate the intricacies of financial crime in a digital age today.

In recent years, the landscape has shifted. Financial institutions are doing their best to move from reactive to proactive monitoring. At the heart of this transformation is artificial intelligence and machine learning (ML/AI). ML/AI allows systems to analyze transactional data, behavioral patterns, customer profiles, detect fraud networks and external risk indicators in real time. The result is earlier, more accurate detection of anomalies—before they escalate into reportable offenses.

Another breakthrough we’ve achieved as an AML community is the Compliance Convergence—the merging of AML, data protection, cybersecurity, and other compliance domains. Operators, increasingly frustrated by the need to manage multiple internal and third-party systems amid shrinking compliance budgets, are now seeking a single source of truth—one platform or a well-integrated system—to store customer and transactional data, assess AML risk, make decisions, and manage the entire compliance lifecycle.

Today, compliance and AML officers are eager to have customer due diligence (CDD) data feed directly into transaction monitoring processes, which allows systems to inform necessary follow-up measures and, ideally, initiate them automatically. For example, if there’s concern about the device used to log in or if a customer is transacting at unusual hours, well-established controls can be triggered—even before a transaction is approved—when supported by the right technology.

Bad actors are often one step ahead, so when friction is introduced through transaction monitoring, they quickly adapt with more intelligent evasion tactics. To counter this, modern systems are now designed with threshold, frequency, geographic, pattern-based, and behavioral rules in mind. The standard today is far more holistic—it enables more adaptive and effective detection than ever before. New functionality is now being introduced into existing transaction monitoring systems, with the growing use of AI for network and anomaly detection taking center stage. Software vendors are carefully designing what “good” looks like while remaining mindful that each operation is unique and system flexibility is essential.

In the UK, for example, regulatory sentiment is that of acknowledgment and of warning; the FCA in its Money Laundering and Terrorist Financing Guide provides examples of good practice which include the exploration of new approaches to automated monitoring (e.g., network analysis or machine learning).

Over time, we’re seeing a shift in how financial crime-related issues are viewed—driven by technological advancements, cost and operational considerations, stronger policies and regulations, and changes in business processes.

Adaptive systems have evolved in step with emerging threats. Unlike traditional rule sets, AI-powered engines continuously learn from new data, refining detection models and uncovering laundering methods without manual input. This adaptability has significantly reduced false positives, freeing up resources and improving compliance efficiency.

Real-time monitoring is now essential. As transaction volumes and speed increase, institutions analyze activity instantly across multiple channels, enabling immediate risk responses. Natural language processing (NLP) helps interpret unstructured data—like transaction memos or communications—revealing hidden links and suspicious patterns that legacy systems used to miss.

Another major shift is toward risk-based, customer-centric monitoring. Institutions are moving beyond transaction-level scrutiny toward 360-degree customer views, combining KYC data, transactional history, device telemetry, and external sources (e.g., news, social media) to create dynamic risk scores. Regulators now expect systems to be tailored to individual behavior, product use, and geography—boosting both detection accuracy and operational efficiency.

Sanctions and PEP screening have also improved. AI tools now detect aliases and cross-border ties more effectively, reducing onboarding delays and enhancing ongoing oversight.

Looking ahead, the future of transaction monitoring lies in deeper data integration, predictive modeling, enhanced automation, and stronger collaboration across compliance, cybersecurity, and regulatory bodies, as well as the end-to-end compliance enablement—One Platform to meet all compliance needs. The goal is clear: to stay ahead of financial crime with intelligence-driven, adaptive systems.

For institutions investing in these capabilities, the payoff extends beyond compliance—toward a more resilient, future-ready defense.

Natalie Buraimoh

Head of AML Product at Sumsub