When discussing the need for smarter crime detection techniques, Julie Conroy, research director at the Aite Group, does not mince words.
“Between two and five percent of global GDP is currently being laundered, and with legacy approaches we’re catching less than one percent of it,” she says.
IBM recently hosted a set of virtual panel discussions on the subject of financial crimes, where panelists discussed the exciting new AI tools that banks and other financial institutions are using to combat money laundering, tax evasion, drug and human trafficking, and a host of other criminal activities.
An emerging theme was the convergence of Know Your Customer (KYC) practices, which usually occur when a new customer is being onboarded, and Anti-Money Laundering (AML) practices, which continue throughout the customer life cycle. KYC typically provides a snapshot of a customer’s risk at a specific point in time. That data, in the past, was typically siloed separately from their behavior as a customer. This fragmented approach is vulnerable to fraud and prevents institutions from maintaining an accurate, dynamic portrait of a customer, limiting their ability to provide the highest level of service.
A new OEM agreement between IBM and leading Client Life Cycle Management (CLM) provider Fenergo actualizes the convergence of KYC and ALM.
“Data is at the center of this,” says Fenergo CEO Marc Murphy. “Banks need to know their customer. They need to know the behavior of their customer. All of that is underpinned by data.” The good news for fintech startups and neobanks is that usually, they have already put data collection and analysis at the forefront.
Through the OEM agreement, IBM and Fenergo can create solutions that combine Fenergo’s CLM offering with the IBM RegTech portfolio of AML and KYC solutions, all built with artificial intelligence from Watson. As a result, IBM plans to offer clients an AI application suite focused on risk and compliance.
IBM RegTech offers an innovative approach, which merges KYC and AML and super-charges them with AI tools like Natural Language Processing, Robotic Process Automation, and Machine Learning. This powerful combination improves prediction accuracy and reduces the cost of human management. While the financial industry has gradually moved toward automation for years, the current COVID-19 pandemic and its associated wave of scams has accelerated the need for banks to step up their financial crimes detection capabilities.
Conroy says that the sudden rush of government stimulus money will result in a burst of fraudulent activities. She’s seen an uptick in application fraud attacks on the lending side, and mule account applications preying on unemployed people.
“Scammers thrive on chaos and confusion,” she says, citing upheavals like the current mass workforce migration. “Once you’ve done your KYC, you need dynamic segmentation and other tools to look at the ongoing activity of the business. That’s going to be key to catching some of the fraud that’s flooding the system.”
“What we’re trying to do as an industry,” says Austin Wells of IBM’s Financial Crimes Insight team, “is understand the potential risk that each entity and group of entities poses to the financial institution from the first time we see them to the last time we see them.”
Wells describes two problems that prevented this shift toward automated solutions from appearing until now. The first was a technology bottleneck. Finding, collecting, normalizing, and parsing data in order to extract meaning from it requires lots of computing and a robust cloud infrastructure, which has only recently been possible. The second was a lack of trust on the part of operators and regulators in the ability of AI to make accurate assessments and predictions. But as AI technology has improved with measures to mitigate data drift and bias, and has now been proven out in the marketplace, regulators and financial institutions are now ready to embrace the inevitable automated future.
IBM’s VP, Global Leader Financial Crime, Clark Frogley, echoes this sentiment: “Gone are the days of the black box,” he says. “You need supporting evidence and documentation. You can’t just say ‘The model told me that’s what we needed to do.’” Transparency and explainability are imperative, and they must advance apace with our ability to ingest vast quantities of data.
IBM Fellow Donna Dillenberger adds a few more promising AI technologies to the list. We now have AI with anti-bias capabilities. The latest toolsets are able to use AI learning techniques in order to decide the most appropriate types of AI models to apply to a particular data set, based on the needs of the practitioner. And enterprise AI can help you provide security with Hyperprotect Containers that can guard against attacks on AI models and data. IBM Researchers are even looking as far ahead as quantum computing security, with quantum-safe encryption technology.
Another innovation Dillenberger describes is the use of multi-layer data graphs, which are able to use neural networks with limited data. Not all data needs to be labeled, and the graph learns patterns without human supervision. It scales bigger and learns faster than other graphs in the marketplace.
“For a billion-node graph, we can do graph analytics within 15 minutes. The closest graph platform out there takes hours,” she says.
Multi-layer graphing allows practitioners to find patterns that would have been difficult to ascertain with traditional data analysis methods. As an example, Dillenberger shares a hypothetical scenario where a bank uses AI to track bank transactions, checks it against the COVID-19 news, and then applies a third layer looking into supply chain data. Multi-layer graphing allows the practitioner to track many types of data points at once, and use all of that information to make a more informed decision, all while helping to reduce the level of human analysts needed for the process.
“Many of them stood up their AML and fraud functions as a much more holistic and integrated function,” says Conroy. “Versus the silos that many of the traditional financial institutions still have. That provides them with a more integrated ability to perform both KYC obligations as well as integrating those with a very tight approach to application fraud prevention.” But legacy institutions are catching up — they can’t afford not to.
Banks especially need AI now, according to Michael Dawson, managing director at Promontory: “We’re going to be in a persistently low interest rate environment for some time. It’s going to depress net interest income. Cost control is going to be increasingly important for our clients going forward. To control those costs in a way that meets regulatory expectations around AML will be top priority for many institutions, and AI will be key to bring that about.”
Find out more about Fenergo’s CLM partnership with IBM. Visit ibm.com to book a consultation with a Financial Crimes Insight expert.
IBM Forward-Looking and Cautionary Statements
Except for the historical information and discussions contained herein, statements contained in this release may constitute forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. Forward-looking statements are based on the company’s current assumptions regarding future business and financial performance. These statements involve a number of risks, uncertainties and other factors that could cause actual results to differ materially, including the following: a downturn in economic environment and client spending budgets; the company’s failure to meet growth and productivity objectives; a failure of the company’s innovation initiatives; damage to the company’s reputation; risks from investing in growth opportunities; failure of the company’s intellectual property portfolio to prevent competitive offerings and the failure of the company to obtain necessary licenses; cybersecurity and data privacy considerations; fluctuations in financial results, impact of local legal, economic, political and health conditions; adverse effects from environmental matters, tax matters and the company’s pension plans; ineffective internal controls; the company’s use of accounting estimates; the company’s ability to attract and retain key employees and its reliance on critical skills; impacts of relationships with critical suppliers; product quality issues; impacts of business with government clients; currency fluctuations and customer financing risks; impact of changes in market liquidity conditions and customer credit risk on receivables; reliance on third party distribution channels and ecosystems; the company’s ability to successfully manage acquisitions, alliances and dispositions; risks from legal proceedings; risk factors related to IBM securities; and other risks, uncertainties and factors discussed in the company’s Form 10-Qs, Form 10-K and in the company’s other filings with the U.S. Securities and Exchange Commission (SEC) or in materials incorporated therein by reference. Any forward-looking statement in this release speaks only as of the date on which it is made. The company assumes no obligation to update or revise any forward-looking statements.