How Artificial Intelligence can detect and combat money laundering networks

AI technology has proven to be reliable, especially when it comes to detecting money laundering, and is empowering leading firms to tackle the issues in an increasingly effective manner, writes John Spooner.

Artificial Intelligence (AI) has evolved from being a technology buzzword to the commercial reality it is today.  This technology is making a positive impact across many industries, including the financial sector.  

The financial services industry has a reputation for constant innovation in the pursuit of finding new revenue opportunities. This is happening across all segments including capital markets, commercial banking, consumer finance and insurance.

The use of AI in financial services is changing the business landscape, even in traditionally conservative areas.  According to a Bank of England survey of 500 UK financial institutions, two-thirds of respondents were reported to have already been using machine learning in some form, with the median firm using live ML applications in two business areas.  This is expected to more than double within the next three years. 

Financial institutions today utilise AI for areas such as customer service, risk management, fraud detection and anti-money laundering, while adhering to regulatory compliance. AI technology has proven to be reliable, especially when it comes to detecting money laundering, and is empowering leading firms to tackle the issues in an increasingly effective manner.

Traditional approaches for tackling money laundering

Money laundering poses a serious threat to the financial services sector.  Fines for banks that fail to prevent money laundering are now worth more than $10 billion per year.  Banks have constructed large teams, and allocated them time-consuming tasks of identifying and investigating suspicious transactions, which often takes place within a complex network of players.

Typically, investigation teams use rule-based systems to identify any suspicious transactions. This rule-based workflow consists of the following three steps:  Firstly, an alert is generated by the alerting system; secondly, the investigator reviews it using information from different sources and finally, the alert is approved as True Positive or classified as False Positive.  

A False Positive can be defined as an error in data reporting, in which a test result improperly indicates the presence of a condition that in reality is not present.

However, the problem with rule-based systems is that they create a large number of false positives, usually in the range of 75 to 99 percent.  These means that a vast amount of time and manual effort is being wasted to investigate these false alerts.  The high number occurs because the rules can become outdated quickly and it takes time for the systems to be re-coded.

How AI can address false positives

Anti-Money Laundering (AML) programmes deploy rule-based transaction monitoring systems, spanning areas across monetary thresholds and money laundering patterns. However, bad actors can adapt to these rules over time, and tweak their methods accordingly to avoid detection. 

This is where AI-based behavioural modellingand customer segmentation can be more effective by discovering transaction behaviours and identify behavioural patterns and outliers that indicates any potential laundering.

AI, especially time series modelling, is particularly effective at examining a series of complex transactions and finding anomalies.  Anti-money laundering using machine learning techniques are able to identify suspicious transactions, and also irregular networks of transactions. These transactions are flagged for investigation, and can be scored as high, medium, or low priority, so that the investigator is able to prioritise their efforts.

AI can also provide reason codes for the decision to flag transactions. These reason codes tell the investigator where they might need to search to uncover the issues, and help to streamline the investigative process.  AI is also able to learn from the investigators throughout the review, clearing any suspicious transactions and automatically reinforcing the AI model’s understanding and ability to avoid patterns that don’t lead to laundered money.

AI vs. rule-based systems

AI-powered AML systems provide many advantages over an existing rule-based system.  This includes being able to dramatically reduce false positives, provide a curated set of alerts to the investigator and the ability to ingest domain-specific IP customised for money laundering.  

The AI technology can be strategically placed between the AML rule-based system and the investigator, which allows companies to gain a rapid return of investment.  Overall, the average investigation time is dramatically reduced from between 45 to 90 days to mere seconds. It also greatly reduces any human inaccuracies and hours required per person, and can fit rule-gaps with innovative features.

Address money laundering and drive productivity

When used effectively, Artificial Intelligence (AI) can be a critical factor to success in the financial services industry. It enables financial services companies to not only efficiently build personalised banking experiences, fraud and money laundering models but will also improve employee and business productivity.  

As money laundering networks become ever more complex, the time is now, for progressive financial intuitions to start embracing AI in order to effectively combat money laundering, and to focus even more effectively on driving overall productivity.

John Spooner is Head of Artificial Intelligence, EMEA at H2O.ai

The views and opinions expressed in this Viewpoint article are solely those of the author(s) and do not reflect the views and opinions of Fintech Bulletin.