My previous post focused on the urgent Know-Your-Customer (KYC) and Anti-Money-Laundering (AML) issues facing financial institutions and the specific needs of business units. Business units are under pressure to keep pace, but so are compliance and auditing. All three lines of defense for KYC and AML must grapple with manual processes and huge volumes of dirty and duplicate data.
Obviously this isn’t ideal. As Former U.S. Deputy Attorney General Paul McNulty famously said: “If you think compliance is expensive, try non-compliance.” The consequences of non-compliance with KYC and AML regulations are crippling. According to Fenergo, in the last decade, penalties amounted to $26 billion globally. And a recent LexisNexis study reports that anti-money laundering (AML) compliance costs US and Canadian financial institutions a whopping $31.5 billion each year.
Effective AML programs are essential for preventing criminal use of the world’s financial networks, and avoiding potentially crippling penalties for financial institutions.
Too Tough to Handle Manually
Faced with an issue too formidable for busy financial services professionals to tackle unassisted, the next question becomes clear: can technology help solve the problem?
In December 2018, the Board of Governors of the Federal Reserve System, FDIC, FinCEN, NCUA, and OCC issued a joint statement to encourage banks to consider, evaluate, and responsibly implement innovative approaches, such as machine learning and AI, to meet their Bank Secrecy Act/Anti-Money Laundering (BSA/AML) compliance obligations in order to further strengthen the financial system against illicit financial activity.
According to LexisNexis, machine learning (ML) and artificial intelligence (AI) are expected to become standard for AML compliance processes within five years. Despite strong urging from government agencies and predictions of wide-spread AI/ML adoption, LexisNexis finds that only 25% of financial institutions have adopted AI/ML solutions as of 2019.
Compliance is Overwhelmed
Consider: Among their many other tasks, compliance professionals are faced with government mandates to review a snowballing number of suspicious activity reports (SARs). According to statistics issued by FinCEN, banks alone filed over 700,000 money-laundering-related reports in 2016, an exponentially higher number than in 93,545 filed in 2012. As the numbers continue to skyrocket, forcing compliance professionals to skip steps in clearing SARs, either by not having time to review them, or by creating ineffective alert monitoring rules.
The Solution: With AI/ML systems like Tamr, data related to transactions and SARs can be unified in one place and normalized to create a clean master data set that can be sorted by criteria such as source, account and type of suspicious activity. This information can be augmented with external and internal data sources to conduct more comprehensive transaction monitoring.
What’s important to note here is that Tamr’s machine-driven, human-in-the-loop approach, helps compliance resolve alerts faster by directing ML algorithms to do the heavy lifting. Analyses can be further advanced by incorporating additional reference data to get a more complete view of customers, entities and transactions. Compliance users can even create ad hoc and customizable risk ratings that are able to incorporate additional internal and external sources to assist with specific searches.
The Result: Compliance can review all transactions, SARs and entities in a comprehensive, timely way.
The Impact for Internal Auditing
Consider: On the long to-do list for internal auditors is the overwhelming job of ensuring that KYC programs are based on complete, correct information, that are largely based on customer onboarding questionnaires. It’s especially tough because the task involves dirty, duplicate data. This process normally involves manual review of sample of customers based on specific rules.
The Solution: Tamr unifies, cleans and normalizes customer data, giving auditors a 360-degree view of all their customers that is specifically tailored for auditing purposes. Tamr then enables creation of custom transformations to double-check whether or not every customer has all records properly populated, while checking for small inconsistencies (e.g. customers recorded being from “Iran” instead of “Ireland”). With all the data arranged and customized, auditors can easily cluster all higher-risk customers for further review based on requirements of each specific audit, while pre-clearing others. Like compliance, auditors can generate customized risk ratings that incorporate additional reference data and auditors’ own expertise to help them identify networks of suspicious activity. Testing theories to identify foreign correspondent banks has never been more intuitive.
The Result: Auditors can review an entire population of customers instead of a sample to improve KYC accuracy, completeness and detail.
Upscaling Compliance and Auditing with AI/ML
With multi-billion-dollar fines being imposed against institutions that have run afoul of KYC and AML regulations, it’s time for firms to aggressively adopt AI/ML solutions.
I’ve spent years working in KYC and AML, so, like the famous Farmers Insurance commercial, “I know a thing or two, because I’ve seen a thing or two.” I have always been a proponent of implementing sophisticated models and tools, but I noticed that the general attitude is that AI is going to replace humans completely.
Compliance and auditing professionals urgently need AI/ML solutions—not to replace, but to complement their efforts. Right now, AI and ML are perfect tools in the hands of professionals. They can handle the grunt work, so humans have more time to do exciting and creative work. Only then can financial institutions guard against criminal financial activity and avoid staggering penalties.