Money laundering is a vague crime. This is not because it is legally difficult to name, but because it can rarely be identified in a single process. An account is opened, a company changes its beneficial owner, money moves across several countries. All of these are completely normal processes that are commonplace in economic life.
Modern anti-money laundering measures therefore need to recognize complex patterns. It must capture processes that are deliberately distributed, obscured and accelerated. With AMLD6 and the new European supervision, the European Union is addressing precisely this changed risk situation.
Why Europe is reorganizing supervision
The previous fight against money laundering in Europe had a strong national focus. Each member state implemented European directives into national law, developed its own testing procedures and set up its own supervisory structures. Since national markets function differently, this certainly had advantages. However, it also created gaps because criminal financial flows do not adhere to jurisdictional boundaries.
With the Implementation of AMLD6 An attempt is therefore being made at European level to form a denser system from many individual answers. This affects risk analyses, internal training, reporting channels, technical control systems and the question of how organizations even recognize suspicious patterns. However, new rules can only be effective if they are translated into reliable routines.
A set of rules with multiple levels
AMLD6 does not stand alone, but the directive is part of a larger European money laundering package, which also includes a directly applicable regulation and the new money laundering supervisory authority AMLA.
While a directive must be transposed into national law, a regulation applies directly and therefore ensures greater uniformity. The money laundering package is changing the architecture of regulation. Previous EU directives gave the member states more leeway, but the new package brings certain basic rules closer to the European level. However, national authorities remain important, for example when it comes to specific supervision, sanctions and cooperation with the Financial Intelligence Units, i.e. the national central offices for financial transaction investigations. The result is not a completely centralized system, but rather a network.
AMLA as an institutional break
The new one European supervisory authority AMLA is perhaps the most visible part of the change. Their task is not only to exercise additional control, but also to coordinate supervision, align standards and pay greater attention to particularly high-risk actors. The EU is thus responding to a well-known problem: a cross-border financial market is difficult to monitor if supervision remains organized primarily along national lines.
From a scientific perspective, the institutional logic is interesting. AMLA stands for the attempt to better bring together scattered information. Not every national authority sees the same patterns. A European body can compare abnormalities differently, which promises more overview but also increases the demands on data quality and collaboration.
Data analysis as an early warning system
Modern financial systems produce enormous amounts of data, which are both an opportunity and a burden for combating money laundering. Suspicions often arise not from a single transaction, but from patterns of origin, time, participants and payment methods. A single compliance department cannot meaningfully review millions of transactions by hand. Automated systems, on the other hand, can look for patterns, highlight deviations and make relationships between accounts, people or companies visible.
Since money laundering often takes place through many intermediate stations, methods that analyze networks are particularly relevant. In this way, relationships can be represented and processes that deviate from usual behavior can be marked. This doesn’t replace a decision, but it sorts attention, and attention is a scarce resource in complex systems.
What AI can do
In this context, artificial intelligence is often overestimated, but sometimes underestimated. It is not a magical tool that detects money laundering before it happens. In many cases, however, it can help to evaluate data sets more quickly and to identify suspicious combinations that would be lost in classic rule models.
Rule-based systems often work with fixed threshold values. A payment above a certain amount triggers an audit. AI-supported models can work more flexibly. You can observe behavior over time, compare similar cases or evaluate unstructured information from media, registers and documents.
But this shifts the problem. The question is then not just whether a system finds hits. It also explains why it finds a match.
The limit of automation
Traceability is important in data-based financial supervision. Suspicion must remain justifiable for both legal and scientific reasons. An automatic model must therefore reveal its decision-making processes.
The quality of the input data is also very important. Incorrect registry information, outdated customer data or incomplete transaction information can lead to poor results even with good models. Automation can even accelerate errors if incorrect information is processed without checking and incorrect assumptions are reproduced again and again. Human evaluation therefore remains an important component.
More supervision does not automatically mean more knowledge
The new European money laundering architecture is often described as tightening. This is generally true, but it falls short. More rules, more data and more authorities do not automatically make prevention better. What matters is whether the information fits together.
Registers can create transparency if they are up-to-date, verifiable and linked. But they can also create a false sense of security if straw people, nested chains of ownership or foreign constructions are not recognized. The same applies to sanctions lists and politically exposed persons. Combating money laundering is therefore also a knowledge problem. She asks when data carries enough meaning to trigger action.
The practical test in the organizations
Banks, financial service providers, real estate players, crypto service providers and other obligated parties must translate the specifications into work processes. Only then will it be decided whether AMLD6 and the new European supervision only create a new documentation burden or actually enable more precise control.
Roles and responsibilities are important: money laundering officers need access to information, but also institutional weight. Only if internal reporting systems work without every anomaly being indiscriminately dramatized can a reliable review process be created from individual reports and suspicions can be prioritized sensibly.
Supervision as a learning system
The future of anti-money laundering will probably be shaped less by individual thresholds than by the ability to recognize new patterns early on. Criminal actors are adapting. They use digital payment channels, cross-border structures and industries in which ownership or asset flows are difficult to understand. Regulation always lags behind this change.
AMLD6, AMLA and data-based analysis methods are an attempt to narrow this gap. It cannot be closed completely. Financial systems are too flexible and abuse strategies too inventive for this.
This is precisely why the scientific perspective is helpful. He sees anti-money laundering not just as a list of duties, but as a dynamic control system whose quality depends on how well law, technology, organization and human judgment work together.