Detecting and Preventing Financial Crimes in the Digital Age

Artificial intelligence and machine learning are impacting our world in a myriad of ways. According to a research report published by Accenture, if the 12 developed countries highlighted in its report implement AI and its associated technologies now, they could double their annual economic growth rates by 2035. The impact of AI on business is projected to increase productivity by up to 40%, simply by making workers and workflow processes more efficient. What does this mean for you? AI and machine learning tools now offer credit unions huge efficiencies that are simply not available with the legacy technology offerings that have traditionally dominated the banking industry. Those credit unions that take the lead in implementing AI and machine learning today will reap massive benefits down the road.

Most Americans are familiar with AI, at least from a peripheral perspective, through consumer environments using Apple's Siri, Amazon's Alexa and Google's Home. Since a recent survey suggests more than 11 million Alexa-enabled Echos and Dots have been sold in the U.S., it seems safe to assume that American consumers would also welcome AI-based solutions implemented by credit unions. These solutions could be services tied to Alexa/Siri-like applications designed to speed up the banking process, or AI and machine learning-based implementations creating new efficiencies that can be passed on to members in the form of lower fees, better ATM access and more. Some members might even see these implementations as a selling point. It wouldn't necessarily matter to them if the use of AI was visible, or even part of the online member experience, so long as the credit union made a public commitment to implementing AI tools.

Machines Learn as They Go

Recommended For You

The term "machine learning" is often used interchangeably with AI, but it is not quite the same thing. AI represents the broader concept of machines being able to carry out tasks in a way that humans would consider smart, while machine learning is an application of AI that gives machines access to vast amounts of data and lets them learn as they go, rather than having a human program them. A good example of machine learning from the consumer world is Google's self-driving car. It features sensors that gather data from nearby objects, in this case other cars, cyclists or pedestrians, and categorizes them based on how they are most likely to behave. From a fraud protection perspective, credit unions can apply machine learning-enabled systems to attack the complex set of rules and regulations that govern money laundering compliance. In fact, online payment powerhouse PayPal is considered an industry leader in implementing machine learning to manage its anti-money laundering programs.

Most AML solutions today rely on legacy technology and, as a result, often produce a high number of false positives. These require a significant amount of manpower to investigate, which can be a challenge to credit unions, as resources are not as abundant as they are at larger institutions. In addition, legacy systems are built with expensive hardware and databases that take a large IT team to manage; even then, the investigative results are sometimes flawed and not delivered in real time.

Machine learning starts in a supervised environment, but eventually becomes unsupervised, for the most part. Over time, machines can independently gain the technical ability to quickly search for suspicious financial activity by reviewing the endless amounts of data generated by an institution. For example, a manual search that might take six months could be handled in a machine learning environment in a matter of minutes. And since these machines learn as they go, money launderers will be forced to change their tactics more often in order to stay ahead of the machines and their learning curves.

Money laundering and other financial crimes aren't going away. We know that credit unions are going to be faced with sifting through an increase in fraud-related data on a daily basis, rather than less. To save money, improve efficiencies and generate better compliance results, it makes perfect sense for credit unions to engage with and implement AI and machine learning tools now. What are some examples of business-critical functions where AI and machine learning can disrupt, enhance and improve? Perhaps one of the most important is in Know Your Customer programs.

A century ago, it was understood that a successful banker would actually know their customers. The home town banker was intimately acquainted with neighbors investing money or seeking loans. Life is obviously just not that simple anymore, particularly with global institutions with hundreds of branch offices in multiple countries. For this reason, KYC compliance is strictly enforced by government agencies; non-compliance comes with stiff financial penalties. KYC is one area where the ability to capture and analyze data quickly and accurately is vital. By incorporating AI and machine learning into risk management programs, credit unions can avoid these penalties.

I don't think we have seen even a slice of the potential efficiencies AI and machine learning tools can bring to the table for credit unions; some big banks have already made deep investments in AI and machine learning. HSBC, which was fined $1.2 billion in a money laundering case five years ago, is now involved in a Google Cloud pilot to implement machine learning to help it stay compliant in the future. How big are the datasets being analyzed by HSBC? In 2014, these datasets clocked in at 56 petabytes, but have since doubled in size to more than 100 petabytes! As more and more banking consumers adopt digital services, the size of the datasets analyzed by big banks, and by medium-sized banks and credit unions, will only get bigger.

Machine learning is here to stay – embrace change and move forward, or peril in its wake.