From monolithic to adaptive: Rethinking watchlist screening in Europe

Across Europe, financial institutions are navigating a compliance environment that grows increasingly more complex. The surge in cross-border payments, coupled with rapid fintech innovation, is bringing these financial institutions into closer alignment with global banking networks and correspondent relationships. But this access hinges on the ability to demonstrate robust anti-money laundering (AML) and sanctions screening capabilities. And for many European financial players, the challenge is magnified by the continent’s rich linguistic diversity and the complex tapestry of overlapping regulatory frameworks, writes Cedric Iggiotti, VP Product, Screening at ThetaRay (pictured, below).

Watchlist screening, which detects transactions involving sanctioned individuals, entities, or jurisdictions, is at the center of this challenge. Yet the systems many institutions still rely on were designed for a simpler era. Monolithic, one-size-fits-all platforms apply the same rules and thresholds across all cases and struggle to keep pace with modern demands. The result is compliance processes that are expensive, inefficient, and in some cases, ineffective.

AI as a regional enabler for smarter screening

It is tempting to present artificial intelligence as the ready-made solution. AI-powered tools, particularly those using machine learning and natural language processing, do offer meaningful advances in name-matching, linguistic coverage, and contextual risk detection. But for practitioners who live under the scrutiny of regulators, the truth is more nuanced than “AI good, legacy bad.” AI in production settings is neither flawless nor universally accepted as a standalone replacement for existing methodologies.

In controlled test environments, a well-trained AI model might indeed recognise that “Altaibi Khaled Aedh G.,” “Khalid Al OTAIBI,” and “خالد العتيبي” refer to the same person. In live, multilingual, and messy data environments, however, such entity resolution remains imperfect. Results depend heavily on preprocessing, normalization, training data quality, and ongoing tuning. Even then, false negatives remain a risk, and in the compliance world, a single missed hit can be far more damaging than a thousand false positives. For that reason, most regulatory-grade deployments now operate as adaptive architectures that combine deterministic rules, fuzzy matching, and AI layers in flexible ways. This allows institutions to benefit from AI’s strengths without losing the transparency and explainability required for audits. 

This adaptive shift is not just a matter of technical preference. It is a regulatory requirement. European supervisors under EBA guidance, BaFin oversight, or the emerging AMLA framework expect immediate sanctions list ingestion, daily customer re-screening, and proof of explainability. The Instant Payments Regulation adds further pressure by requiring funds to be credited within ten seconds. Monolithic systems can hardly meet these parallel demands. Adaptive screening, designed to be context-aware, can. Model risk management disciplines still apply, including documented governance, independent validation, reproducible outcomes, clear audit trails, and strict version control. 

Compliance is more than name matching. It encompasses licensing regimes, beneficial ownership aggregation rules such as OFAC’s 50% threshold, timely ingestion of list updates, structured case management, and the ability to evidence decision-making years after an alert has been cleared.

Linguistic complexity is not an abstract concern. Names across Europe may appear in Roman, Cyrillic, Arabic, or Amharic scripts, with transliteration inconsistencies, phonetic variation, and local conventions such as multi-component patronymics. Transliteration inconsistencies and local naming conventions mean institutions must calibrate screening to multiple scripts and variants. Traditional monolithic systems tend to treat these challenges in the same way, applying uniform thresholds and rules regardless of script, name, culture, entity type, or jurisdiction. Adaptive screening introduces situation-aware strategies. Matching logic can be adjusted for Slavic versus Arabic names, personal names versus corporate identities, or high-risk versus low-risk transactions. By tailoring rules to data quality and risk profile, institutions achieve stronger recall without overwhelming analysts.

Screening in high-volume, mobile-first economies may require processing tens of thousands of transactions per minute. Adding complex machine learning layers can slow throughput if not engineered carefully. Adaptive systems allow institutions to control where AI should augment rather than replace deterministic layers, tuning thresholds based on risk exposure and appetite. 

From regulatory burden to strategic advantage

The case of Travelex, which operates across Europe, illustrates the potential impact of this approach. By introducing adaptive, AI-driven screening, the company reportedly cut false positives up to tenfold in some of the jurisdictions they operate. But such gains depend on more than the AI engine itself. When done well, adaptive watchlist screening can shift compliance from a perceived cost center to a strategic capability. Faster onboarding, reduced analyst workload, and more precise alerts build customer trust and open doors to new partnerships. But this is not a turnkey solution. It is an investment in governance, people, and processes that must be sustained long after deployment.

The institutions that will set the standard in Europe are those that combine technological ambition with regulatory discipline. Those that acknowledge AI’s boundaries as openly as its strengths, reference sanctions regimes accurately, and resist the temptation to reduce a complex, regulated discipline to a few oversimplified claims. In sanctions compliance, credibility is built in the details, and the future will belong to those who move beyond monolithic systems and embrace adaptive architectures that can deliver speed, accuracy, and transparency in an increasingly demanding regulatory landscape. 

Cedric Iggiotti is VP Product, Screening at ThetaRay, where he drives product strategy in sanctions and entity screening. He was formerly CEO and co-founder of Screena, a specialist screening startup now integrated with ThetaRay.

Share this article: