Interview: Frans Thierens on AI and how it reduces the complexity of modern anti-money laundering

Modern technologies are becoming more sophisticated, but so are modern money launderers and the organisations involved in financing international terrorism. In this conversation, Frans Thierens, Anti-Money Laundering Compliance Officer (AMLCO) at KBC Bank, digs into the advances and new complexities in anti-money laundering legislation. His responses highlight the important role of AI technology in identifying illicit transactions and suspicious patterns.

Frans, would you first walk us through the essence of transaction monitoring obligations in anti-money laundering (AML) regulation?

The primary focus of AML legislation is to detect atypical transactions that cannot be matched with the customer’s profile.

Belgian and European laws place the duty of vigilance on front offices: the people in direct contact with customers. In the context of a digital world, though, human alertness should be supplemented with automated monitoring systems, which can assist with identifying atypical transactions based on various considerations.

Traditionally, rule-based systems generate alerts based on parameters such as limits – for example, transactions coming from high-risk countries and exceeding a predetermined amount. However, those systems do not allow for profiling based on comparison with peers or other behaviors. Artificial intelligence (AI) can identify behaviors from large data sets and assign scores to customers, indicating their likelihood of involvement in money laundering.

Can you elaborate on KBC’s use of AI in AML monitoring? Why and how does AI come into play?

For now, KBC uses AI to augment and support rule-based systems, which has increased efficiency and effectiveness. AI can also identify complex patterns, such as social fraud, by analysing additional features as a targeted approach to suspicious activities.

Profiling involves comparing a customer’s transactions to their previous behavior and to the behavior of its peer group in order to identify variations and explain atypical transactions, like sudden international money transfers due to a job relocation. Employing an AI model to evaluate customer behavior fits within the spirit of the legislation.

Local regulators focus on timely reporting of transactions. Ex ante AML processing often clashes with the digital habits and possibilities offered to clients as well as with rapid processing requirements, like the European Payment Services Directive (PSD). Ex ante monitoring often targets fraud, requiring swift action, while money laundering activities are continuous and their complex patterns can only be observed over time.

How have money laundering activities changed over time? How can AI help combat these newer, more sophisticated approaches?

The money mules of today are no longer selected but rather created. In the past, people like students, unemployed individuals, notaries and doctors were persuaded to make their accounts available to receive funds and act on the instructions of the money launderer or fraudster.

When criminal enterprises “create” money mules, they essentially fabricate elaborate artificial digital identities from scratch.

These mule accounts often involve the creation of a synthetic identity, built from real but stolen identities combined with entirely fake data. Mules can also be fully automated shell personas without any authentic human information or action behind them at all.

For example, a fraudster creates a fake identity, then opens a bank or crypto account in that name, using it to funnel money from scams or cybercrime operations. This continues for as long as the account remains uncompromised before switching to the next.

With this type of complex and sophisticated digital approach, money launderers have identified ways to reduce the risk of exposure or non-compliance that persuaded or coerced human mules may have presented in the past.

This new approach is further complicated by a combination of legal and illegal activities. AI technology can enhance the process of investigating dormant accounts and identifying patterns, such as for example multiple customers wiring funds to the same gold dealer.

Payments to social security, tax authorities, salaries and gold dealers may appear legal at a glance, but they require deeper investigation to reveal concerning patterns.

Due to data restrictions, visibility is limited to the flow of data within one’s entity, so activities at other institutions or in other countries add complexity to the process. The human mind is creative, but it unfortunately cannot capture every important detail from memory.

Added experience from investigators and insights from sources like the Financial Intelligence Unit (FIU) and National Bank will help to identify suspicious behavior.

Frans Thierens Anti-Money Laundering Compliance Officer (AMLCO) at KBC Bank

Moving forward, how do you anticipate AI will transform AML monitoring?

AI is currently limited to its own data, but an overarching system could identify patterns across institutions without violating customer confidentiality. Added experience from investigators and insights from sources like the Financial Intelligence Unit (FIU) and National Bank will help to identify suspicious behavior and further tighten the net around the professionalised organisations engaging in money laundering as a service.

The EU Anti-Money Laundering Authority (AMLA) aims to centralise intelligence in Frankfurt in an effort to Europeanise national systems. While on an institutional level KBC lacks visibility on transactions at other financial institutions, Belgium’s FIU has the authority to connect suspicious activities across multiple banks within the country. And at the European level, AMLA can require system access and overcome technical challenges to consolidate suspicious activities across borders.

Frankfurt was chosen from a list of 9 candidates to become the enforcement hub for the EU’s fight against money laundering.

The German city was selected through a series of public hearings, which granted the new Frankfurt-based AMLA direct and indirect supervisory powers over obliged entities and empowered the Authority to impose sanctions and measures against money launderers and those engaged in the financing of terrorism.

Frankfurt is of course home to the European Central Bank and has long been a well-established international financial hub within Europe. In addition, local and national governments pledged their financial, logistical and political support. As such, Frankfurt was deemed to have the appropriate talent, logistics and infrastructure already in place to be able to spearhead critical initiatives, coordinate with local and international authorities, and facilitate anti-money laundering efforts at the highest level.

What impact do laws and regulations have on the AML process?

Regulations, especially the GDPR, can admittedly complicate processes. In Belgium, tax authorities fought for years to gain visibility into accounts. Combining customer databases with transaction databases is complex. In the Netherlands, for example, GDPR hindered progress on a modest setup.

Article 75 of the new EU AML package offers hope, but it does require agreement from local regulators and data protection authorities. The concept is promising in theory but demands significant practical work.

Embedded in the EU’s Anti-Money Laundering Regulation (AMLR), Article 75 details how financial institutions and other obliged entities may form supervised partnerships. Its purpose is the ability to share customer information with the specific goal of combatting money laundering and the financing of terrorism.

Information sharing is permitted only for higher-risk cases or where further assessment is needed to determine risk.

Supervisory authorities must pre-approve partnerships, ensure data protection compliance, and may consult FIUs and data protection agencies. Only specified categories of customer and transaction data may be shared, and participating entities must maintain internal policies and records. The provision aims to support targeted, cross-border collaboration while remaining compliant with legal and privacy safeguards.

Frans Thierens is Anti-Money Laundering Compliance Officer (AMLCO) at KBC Bank. With a legal background and extensive training in financial crime compliance, he has been active in the field for multiple decades.