Special contributor report: Transforming KYC/AML using artificial intelligence
Thursday, July 6, 2017
Posted by: Brian Monroe
By Deepak Amirtha Raj
A research analyst for Cenza, specializing in strategy development and business analytics through artificial intelligence, virtual reality and augmented reality.
July 6, 2017
*Special contributor report*
First published here. Republished with permission and appreciation.
Most people would have heard of the headline-making tremendous achievements in artificial intelligence (AI): Systems defeating world champions in board games like GO and winning quiz shows.
These are small realizations of AI, but there is a silent revolution taking place in other areas, including Regulatory Compliance in Financial Services.
"Traditional rule-based KYC-AML technology necessitates significant dependence on manual efforts, particularly in the alert investigation stage, which is costly, error-prone, and inefficient"
The ultimate aim of any Financial Institution (FI) is to earn the confidence and faith of their customers, but equally important is to verify the information customers provide back to them.
That is vital as the accuracy of a bank’s customer due diligence initiatives have a direct correlation to the related customer risk assessments, typically a numeric score that falls into low, medium or high-risk categories. Those values then inform the AML transaction monitoring system, which is typically tuned to give higher risk entities more scrutiny.
The regulators are increasingly concentrating on ensuring that banks have robust and effective controls in place for customer due diligence (CDD). Lapses in money laundering and CDD controls have exposed several FIs to monetary penalties and formal enforcement actions from regulators, resulting in these agencies tightening their levels of supervision on the industry as a whole.
Here are a few examples:
· United Overseas Bank (UOB), which was fined $900,000, and Switzerland's Credit Suisse Group, which was fined $700,000 - both for breaches of anti-money laundering (AML) requirements and control lapses.
· The German lender Deutsche Bank has received a new $204 million fine by British regulator, the Financial Conduct Authority (FCA) for inadequate AML controls.
· The Monetary Authority of Singapore (MAS) has withdrawn the merchant bank status of Switzerland-based Falcon Private Bank's Singapore branch for serious failures in AML controls. In that instance, the severity of the penalty, and individual sanctions against several top officials, came as a result of the bank’s links to the massive 1MDB fraud, corruption and money laundering scandal.
For FIs, CDD is a vital element to de-risk and protect themselves against possible financial crimes.
Driving Changes in KYC|AML
The know your customer (KYC) space faces several key driving changes that must be also taken into account before implementing a CDD program aided by AI:
1) Ever-Evolving Regulations - Regulatory requirements and compliance environments, informed by initiatives like the tax-centric U.S. Foreign Account Tax Compliance Act (FATCA), aggressive sanctions enforcement by the U.S. Treasury’s Office of Foreign Assets Control (OFAC), Europe’s Fourth AML Directive (4AMLD), MAS guidelines can vary across the globe considerably. As a result, multinational banks need to ensure compliance not only in their home country, but also in environments that are more complex and have fewer regulatory infrastructures.
2) More Regulatory Scrutiny - Not adhering to the financial crime compliance laws in any given jurisdiction can have very punitive outcomes for the organization. For Example, Deloitte has highlighted in its “Meeting new expectation” report that AML sanctions-related fines and penalties imposed in 2013 and 2014 quadrupled the total for the previous nine years.
3) Cost Pressure - An average bank spends £40 million a year on KYC compliance, according to a recent Thomson Reuters survey, which also revealed that some banks spend up to £300 million annually on KYC Compliance and CDD. Not surprisingly, to counter increasingly creative criminals, satisfy ever more critical regulators and get valuable intelligence to law enforcement, some feel that the cost of compliance is increasing exponentially.
4) Legacy Systems - The KYC documents are retained over several years in document management systems, these are not always easy to locate. At that same vein, it can be wildly expensive for banks to decide to remediate and re-risk rate their entire customer populations, an initiative that, while bringing the potential to uncover suspicious customers who have fallen through the cracks, could irk rank-and-file clients.
Artificial Intelligence in KYC AML
Artificial Intelligence (AI) takes KYC and AML compliance to the next level.
AI isn't just a technology, it is a collection of related technologies which has the potential to automate workflows and quickly analyze large volumes and different types of data. Some of the potential benefits of using AI in KYC/AML are:
1. Link Analysis:
AI-based link analysis is a set of techniques for exploring associations among large numbers of objects of different types. These methods are crucial in assisting human investigators in comprehending complex webs of evidence and drawing conclusions that are not apparent from any single piece of information.
These methods are equally useful for creating variables that can be combined with structured data sources to improve automated decision-making processes. Typically, linkage data is modeled as a graph, with nodes representing entities of interest and links representing relationships or transactions along with dubious jurisdictions, companies, ultimate beneficial ownership (UBOs) and other details.
“AI-enabled solutions can not only automate significant parts of operations but also offer superior insights through advanced capabilities for analyzing structured and unstructured data.”
2. Pattern recognition:
In most cases, money launderers hide their actions through a series of steps, making money derived from illegal or unethical sources appear to have been earned legitimately.
As a result, and to lower the number of false positives AML transaction monitoring analysts have to sift through, most of the major banks across the globe are shifting from rule-based software systems to AI-based systems, which are more robust and intelligent to uncover a broader array of potentially illicit AML transaction monitoring patterns.
Recently FICO has developed Anti-Financial Crime Solutions which uses unsupervised Bayesian learning techniques to understand customer behavior, which is further used to drive investigations and possible SAR filings, just as an example.
3. Unstructured Data Analysis:
AI in KYC relies more on Natural Language Processing (NLP) and supervised machine learning techniques.
Each of these technologies has specific uses and NLP, in particular, is starting to come into widespread use in helping to analyze unstructured content, such as adverse media, also called negative news.
Together with machine learning, NLP-based AI can "read" such articles and perform a range of tasks, including extracting metadata, identifying entities that are referred to, and "understanding" the intent or purpose of specific parts of the document.
4. Workflow Automation:
AI can also be used in generating documents, reports, audit trails and notifications.
For instance, in analysis by cognitive computer platform DDIQ’s, reports generate risk profiles on both companies and individuals in just minutes, providing comprehensive and in-depth global due diligence information.
Also, the reports provide links to the data sources, enabling them to be fully auditable, vital details for internal audit teams and regulatory examiners who typically want to know the accuracy, veracity and origin of any information used in AML decision-making.
This capability is even more critical as recent and upcoming changes to global KYC regulations will require the identification of and due diligence on beneficial owners.
What do the AI providers think of their prospects?
Naturally, they are optimistic about the potential to win bank clientele. "The catalyst for widespread adoption of AI for KYC/AML tasks will likely be the concern about competitive edge," says David McLaughlin, chief executive of QuantaVerse based in Wayne, Penn., in a statement to FinOPs last year. “As is the case with all new technology, once a few large players become interested, the rest fall in line.”
Regardless of all the new technology, AML professionals should not worry about losing their jobs anytime soon. That’s because, currently, the prevailing sentiment is that there aren't enough qualified AML professionals to go around.
With regulators only too eager to penalize banks for any failures in KYC/AML compliance, institutions are staffing up.
“Al doesn't replace human intelligence, but improves it,” said Mallinath Sengupta, chief executive of NextAngles, a subsidiary of Mphasis Corp, in a discussion on the issue in September of 2016.
He told FinOps. “It's true, banks will continue to use analysts to make the ultimate decisions on whether a transaction is suspicious and must be reported, but they can be more productive and have more confidence that their decision is accurate.”
About the author
Deepak is a researcher, strategist, and writer. He studied engineering and business at Saint Joseph's College. He helps firms by teaching them about Artificial Intelligence, Augmented & Virtual reality, Machine learning and Big Data.