As the EU’s new Anti-Money Laundering Authority raises expectations around AML, firms are moving beyond transaction monitoring towards behavioural intelligence. Adam McLaughlin, Director of Financial Crime at Fenergo, argues that whilst AI will become central to that transition, human judgement remains the deciding factor.
Historically, financial institutions have approached anti-money laundering as a process centred on rules, documentation and transactions. Firms collected customer information during onboarding, screened transactions against sanctions lists and investigated activity that triggered predefined alerts.
The model is no longer adequate. This isn’t because financial crime has dramatically changed; after all, criminals still look for ways to conceal illicit funds and exploit weaknesses in the system. What has changed is geopolitical landscape and the subsequent regulatory response.
Sanctions regimes have significantly grown more complex in recent years, and the opacity of cross-border ownership structures means that establishing who ultimately controls an asset has never been harder. At the same time, regulators expect firms to demonstrate a much deeper understanding of who their customers are and how they behave. This is a key pillar of the EU’s Anti-Money Laundering Regulation (AMLR), which becomes legally binding in July 2027.AMLR introduces a single rulebook and more coordinated supervision across member states. Until now, firms operating across Europe have often had to navigate subtle differences in how individual jurisdictions interpreted the same regulatory principles, creating fragmentation that criminals have historically exploited by targeting weaker points in the system. Greater consistency should significantly reduce that exposure.
Critically, AMLR is backed by a new supervisory body: the Anti-Money Laundering Authority, or AMLA. Based in Frankfurt, AMLA will take direct supervisory responsibility for the highest-risk cross-border financial institutions from 2028 and will act as a central coordinator for national supervisors across the EU. For firms, this means a step-change in supervisory intensity. AMLA will not only set expectations but actively enforce them, and firms that have historically relied on inconsistencies between national regimes will find that latitude significantly reduced.The more significant shift, however, is what regulators now expect firms to detect. For many years, compliance programmes have focused on identifying individual transactions that breach sanctions rules or predefined thresholds. Firms are now expected to demonstrate that they can recognise behaviour that is genuinely unusual or suspicious, requiring a much deeper understanding of customers over time. Criminals often move funds through complex corporate structures and cross-border ownership chains, making audit trails significantly harder to track. Understanding who ultimately owns an asset, and whether their behaviour remains consistent over time, is considerably more challenging than simply completing KYC at onboarding.
The manual process of gathering documentation, reconciling different audit trails and carrying out due diligence is incredibly resource-intensive and time-consuming. This is where technology becomes a genuine enabler. Platforms that combine AI-driven data analysis with workflow automation can process large volumes of information and surface potentially suspicious patterns across multiple data sources in a fraction of the time it would take a human analyst, freeing compliance professionals to focus on genuinely higher-risk cases. AI isn’t going to replace the jobs of compliance professionals. AI can identify anomalies at remarkable speed but determining whether those anomalies genuinely warrant further investigation remains a fundamentally human decision.
There’s a reason why many financial crime teams employ former police officers and investigators. These professionals have accumulated years of experience recognising unusual behaviours and identifying when something simply does not feel right. That instinct isn’t something AI can replicate.
The conversation therefore needs to focus on human-AI collaboration, not one versus the other. Technology will increasingly undertake the heavy lifting, but compliance professionals will still apply judgement, experience and investigative expertise to those insights. The goal is not to automate decisions, but to ensure that human decisions are better informed. AMLA’s arrival reframes what compliance must look like in practice.
For firms operating under AMLR and AMLA’s supervisory oversight, the expectation is that compliance must be a continuous process that builds a deeper, evidence-based understanding of customers over time, rather than a point-in-time exercise completed at onboarding and revisited only when something goes wrong. Achieving that requires firms to combine AI’s ability to process data at scale with the experience and judgement of compliance professionals, using technology to help humans make better decisions instead of replacing them.
Subscribe to our newsletter
