There is as much an art to anti-money laundering (AML) compliance as science.

Sometimes, good compliance boils down to a suspicion by a trained, experienced compliance officer that something is off and requires more investigation. A small anomaly could be indicative of a larger problem. What appears to be an innocent mistake might actually be fraud.

What gets in the way of these moments of clarity, AML experts say, is the cumbersome and manual processes most AML compliance workflows use. So much time is spent sifting through red flags that turn out to be false positives that little time is left for AML compliance professionals to apply their expertise.

In a session at the Association of Certified Anti-Money Laundering Specialists’ (ACAMS) 2022 annual conference held in October in Las Vegas, nearly all the AML professionals in the room reported they spend as much as half their time investigating and eliminating false positives from their workflow. Some reported they spend as much as 75 percent of their time on such tasks.

As currently practiced, AML compliance is inefficient. Compliance professionals need to leverage tools that will eliminate false positives or, at the very least, employ a risk-based evaluation that analyzes which customers, accounts, and transactions warrant a closer look for potentially being involved in money laundering schemes.

While automation has the potential to transform AML compliance, it will not replace human practitioners, experts at the ACAMS conference said. Instead, used wisely, automation tools applying artificial intelligence (AI) and machine learning (ML) will help process flagged accounts and transactions faster, more efficiently, and even more accurately than humans can, freeing them up to focus on the issues that cannot be addressed without human involvement.

“AI/ML can sift through all that information, but it still requires the art of compliance recognizing when something doesn’t make sense,” said Gabriel Hidalgo, managing director at FTI Consulting. “Human intervention is what helps to find the end of the (fraudulent) activity.”

Irene Kenyon, director of risk intelligence for FiveBy Solutions who spent more than a decade as an intelligence analyst, gave an example. She was performing AML due diligence when the name of one LLC caught her attention.

It was called “Behemoth Cat,” which is a character in Russian literature. As a native Ukrainian who speaks Russian, the name aroused her suspicion; eventually, she was able to connect the LLC to a Russian oligarch.

“AI/ML can sift through all that information, but it still requires the art of compliance recognizing when something doesn’t make sense. Human intervention is what helps to find the end of the (fraudulent) activity.”

Gabriel Hidalgo, Managing Director, FTI Consulting

In another case cited by Hidalgo, a money launderer was using the names of Spanish soccer players for LLCs being used to launder illicit proceeds.

“We did not train our AI/ML services to track these names,” he said. “Human interaction draws from your own experience—your gut.”

AI and ML tools can help compliance officers sift through the haystack of information, but that only goes so far.

“Everyone has their own way of following their gut. It’s absolutely critical that (compliance officers) follow it and use it,” Kenyon said. “But you cannot substitute technology for the human element to get to the best end result.”

There has been much written about robots, artificial intelligence, and other computer-powered tools replacing human workers. The World Economic Forum’s Future of Jobs Report 2020 found automation will “transform tasks, jobs, and skills by 2025,” and in that same year, “[T]he time spent on current tasks at work by humans and machines will be equal.”

The same report said while automation might displace 85 million human jobs by 2025, 97 million new jobs might emerge “that are more adapted to the new division of labor between humans, machines, and algorithms.”

This shift toward automation is already happening in AML compliance as more financial institutions employ more complex AI/ML tools to sift through vast amounts of data, searching for trends and patterns that might indicate a money laundering scheme within their network. The tools are also showing promise for ferreting out paths sanctioned Russian oligarchs are using to access the U.S. financial system.

The same experts who said AI/ML tools can aid a compliance professional’s fraud and sanctions investigations cautioned against going too far down an investigatory rabbit hole. If there is a potential crime involved, file a suspicious activity report (SAR) with the Financial Crimes Enforcement Network, as required by the Bank Secrecy Act. But don’t feel like your investigatory skill—burnished by enhanced technology—deputizes you as law enforcement.

“Regulators are looking for reasonable efforts (to identify financial crime), not perfection,” Hidalgo said. “Your SAR is probably not the only SAR to follow this particular activity. Law enforcement will connect the dots using SARs from other institutions.”