Follow the money: Banks need machine learning to fight money laundering

Economic crime is big business; and it’s nasty business whatever way you look at it. The trouble is, laundered money or funds arising from fraud, bribery and corruption, all look much the same as legitimate money. Banks need to employ diligent strategies that help them detect nefarious origins and routes of funds. Deploying machine learning is critical in a battle that’s open on so many fronts, against smart, well-resourced, and often technologically sophisticated adversaries. 

Behind enemy lines

Bad money is in the financial system everywhere. The greater the air of respectability a bank has, the more likely it is to be a target location for the assimilation of ill-gotten gains into the mainstream hustle and bustle, and easily assumable anonymity of a busy economy. Sophisticated illegal organisations of every hue use shell companies, multiple domains, cutting edge software and cryptocurrencies to pull off schemes worth billions.

Money makes the world go round. Unfortunately, it makes parallel, or ‘nether’ worlds go around too; terrorist activities, drug cartels, and organised crime. The fallout from the "world's biggest money-laundering scandal" (Danske Bank/€200bn laundered by Russian entities and others from former Soviet states) reverberates through Denmark, as well as through the global banking industry. “Money laundering problems in a single bank could spread to the entire financial sector and could in turn affect financial stability,” a Nationalbanken report stated. The National Crime Agency (NCA) assesses that “many hundreds of billions of pounds of international criminal money is laundered through UK banks, including their subsidiaries, each year”. (Around ￡14.4 billion, according to one estimate).

The very real threat to the economy, as well as security and safety compromised as a result of the criminal or terrorist activities that the laundered money funds, puts a focus on financial institutions. In some way, shape, or form, these are the conduits through which the money passes.

It’s up to banks to ensure that they stop functioning as ‘safe havens’ for laundered money. As long as the bad guys see banks, and can treat them, this way, reputations are at stake, consumer and stakeholder confidence can be jeopardised and, in a sector that has more than its fair share of pressures – from the entire FinTech/challenger bank scenario, triggering reassessment and overhaul of long-cherished but rapidly outmoded business models, through to an ever-tightening regulatory environment – the long-term damage can be costly at best, and irreparable at worst.

Knowing is good, but knowing everything is better

The threat and the presence of illegal money in the system, and all that it represents, can’t be eradicated overnight, but there are meaningful steps forward that banks often already take in endeavouring to enhance the customer experience that can also be swung into action in trying to degrade the customer experience for certain types of customers (i.e. not fulfilling the function that criminal elements want a bank to fulfil).

Know Your Customer procedures (KYC), for example, are a valuable approach in anti-money laundering strategies. KYC guidelines are designed to prevent financial institutions from being used, intentionally or unintentionally, by criminal elements for money laundering activities. They are the means through which banks verify the identity of account holders to ensure they are not associated with illegal organisations, are not politically exposed persons, and are not linked to criminal activities.

In addition to individuals, banks are also required to Know Your Business (KYB). This is essentially the same process as KYC, but with a focus on identifying a beneficial owner for corporate transactions, i.e. an individual or group of individuals need to be identified in order so that shell companies cannot be used as intermediaries to launder money. KYC and KYB are the twin pillars of an anti-money laundering strategy but their effectiveness depends on one key technology to enable banks to spot the signs of ‘dirty’ money and stop the rot it can cause: a form of Artificial Intelligence known as Machine Learning (ML).

Machine Learning and big data

Major advances in ML are revolutionising fraud prevention and anti-money laundering. Machine Learning offers the financial services sector the opportunity to beat the criminals. By enabling software to successfully predict and react to unfolding scenarios based on previous outcomes, without human input, ML reduces the associated costs and increasing customer satisfaction, and confidence.

ML drives the processing of monumental amounts of transactional and financial data – both customer specific and publicly available – spotting anomalies and detecting patterns suggestive of suspicious behaviour. Rather than looking at each transaction on an individual basis, algorithms process holistically to provide a contextual analysis, instantaneously.

The new generation of algorithms associated with ML learns dynamically, continuously evaluating their performance and adapting to change. This makes it possible to discover new patterns, and keep pace with the changing transactional and threat landscape. Machine Learning can leverage historical data to predict future fraud activities, potentially reducing manual input by almost 50%, and spotting irregularities that evade human scrutiny. The approach enables banks to analyse individual customers and determine whether they or someone else is using their credit card in real time, for example.

Understanding the data that goes in drives an understanding of the results that come out

The caveat with ML is that what you put in dictates what you get out: poor data input produces less than robust/reliable results. As the financial services sector pursues rapid digital transformation, more and more data is created and will then need to be appropriately deployed, balancing the drive to improve fraud prevention with regulatory compliance.

Diverse pressures are coming together at the same time to force banks to reassess how they treat data, and to what degree their existing systems and technology are fit for the purpose of addressing customers’ needs and regulatory requirements at the same time as societal and stakeholder expectations.

Much is made of the role of data analytics in supporting smarter customer experience strategies and outcomes. To take a sideways look at these issues; wherever illegal money comes from, those people are customers too. The big differentiating factor is that they’re not the customers you want. Legally, they’re not the sort of customers you should have anything to do with. So, you have customers you want to delight, and customers you want to fight. In the ongoing struggle against money laundering, ML can help spot the difference. In the process, and in support of KYC/KYB strategies, it provides legal assurance and compliance.

As a final thought, I’d draw attention to the importance of the people factor. ML can’t do it all alone; it is, after all, a ‘machine’ function. Experienced fraud investigators (human ones) are essential in making sense of the lessons and insights that ML produces, in specifying the role it should fulfil, and not least in contributing to the design of technology solutions that enable your bank to fight the good fight in your way.

To continue the conversation or find out more about HPE can help steer your strategy, just get in touch at [email protected].