AI turns financial crime tide | With Kyle Stroombergen

Financial crime and money laundering is the third biggest economic practice in the world and equates to around five percent of the global GDP. Because each country manages its own laws and regulations, the banks are left as the only defense against this attack on the world economy.

The government gave the banks a shot in the arm in the fight against money laundering with the Financial Intelligence Centre Act 38 0f 2001, but FICA makes the banks fully accountable and obliged to comply. That compliance is achieved through gathering the required information and documentation before opening an account. It’s a lot of admin.

“Our users are not data scientists,” says Kyle Stroombergen, head of new deliveries at NGA. The company deploys an artificial intelligence solution for risk-based anti-money laundering called Risk Secure that is smart enough to adapt to changing behavior. Right now the software is web-based and the product has an incredible track record with zero incidences of non-compliance among its clients.

“We developed this product to challenge the traditional screening methods, but it also enhances PEP and sanction lists by adding an extra layer of context,” he explains. “We’ve compiled the world’s biggest adverse media database and clients can be informed in real-time about any corruption allegations.”

NGA Risk Secure is in its seventh generation and bakes the company’s innovative SocialListener AI engine into the process for a comprehensive and detailed assessment widget. By scanning through all media mentions and rating the sentiment, Risk Secure users have near-instant access to a risk report with direct links to the research citations collected from the entire internet.

One-click risk assessment is only part of the story, though. The data providers send FICA lists as PDF and these can be tedious to check against. NGA speeds up this process by using optical character recognition to allow the AI to scan the PDF and cross-reference it with the customer database.

“The concern is always that artificial intelligence will replace human function, but you still need human intervention to apply policies and laws,” says Stroombergen who cut his teeth in database administration and has grown his career alongside this product that is now used in over 35 countries.

Risk Secure began life as a simple, but highly effective name matching software built out of the company’s database expertise. That name matching enabled batch screening which can now process millions of customers a day, even outstripping Oracle for screening volume capacity.

Customer behavior is changing rapidly with the banking experience evolving into a mobile-first model. These consumers demand rapid deployment of real-time assessments to streamline the sign-up process and eliminate barriers to access. Automating the screening and risk assessment frees up time which can be better used analysing and optimising the business functions.

Machine learning allows the AI to cluster banking behaviours to establish a norm for behaviour as customers transition to different banking methods, which in turn exposes anomalies that can be checked by AML officers. This machine learning is supervised, though, but serves as a rapid knowledge tool that further enhances the effectiveness of the AI and can pay those lessons forward to all clients.

“When you partner with NGA you’re bringing 20 years of data science experience onboard,” Strrombergen says. “We operate on an 80/20 rule for customising a product for a particular client because our models have been established over years of service and will only need some input with regards to the client’s internal processes.”

NGA will be moving Risk Secure to a cloud-based system for its full client base and that will come with extra complexity of interfacing with the various cloud provider infrastructure. There is an existing private cloud, however, where clients can access a growing database that has already been categorised and referenced against the regular FICA reporting.

Risk-based anti-money laundering was intended to be a proactive step in the fight against financial crime, but was slowed down by the drag of the human factor. Risk Secure elevates human intervention from the daily slog of database administration to a higher function of policy application and accelerates the data gathering with the power of AI, effectively bringing a gatling gun to a swordfight.        

Leave a Comment

Your email address will not be published.