With new threats and expanding regulations making anti-money laundering processes more complex, financial leaders are turning to AI to help.

“AML programs have evolved over the decades with changing regulations, geopolitical events and emerging financial crime risks,” Jon Glass, a partner in the financial crimes advisory practice at consultancy SolomonEdwards, told FinAi News.

“It all started with the Bank Secrecy Act in 1970,” Glass said, adding that complexity continued to grow with the introduction of suspicious activity reports (SARs) in the 1990s and the 2001 Patriot Act, which mandated formal customer information programs.

“Those evolving regulations have kind of layered on top of each other,” he said. “Geopolitical shocks have repeatedly forced AML programs to adapt.”

Glass said factors driving complexity include:

• The 9/11 attacks, which accelerated global counterterrorist financing rules;
• Escalating sanctions against countries such as Iran, North Korea and Russia;
• A surge in payroll protection and stimulus fraud during the pandemic;
• 2021’s Corporate Transparency Act, which requires certain businesses to file additional reports with the Financial Crimes Enforcement Network (FinCEN); and
• Technology growth, especially AI, which has benefited both FIs and bad actors.

“I view today’s AI moment as the next chapter in this evolution,” Glass said.

Inconsistent, inefficient processes

The biggest bottlenecks in AML operations stem from duplicative and inconsistent manual processes, Glass said.

“Fragmented systems are definitely causing huge bottlenecks, especially when you’ve got banks acquiring many other banks. It takes time to integrate those systems and the workflows and make them uniform,” he said.

Tracy Moore, director of thought leadership and regulatory affairs at fintech Fenergo, told FinAi News that FIs need to prioritize standardizing practices across business segments and departments.

“If you have siloed systems, you may be doing your AML process this way in one set of your organization but are you doing it this other way in a different set of your organization,” Moore said. “The new AI and technology operating models are going to be looking at that more holistically.”

Inconsistent processes might have been a contributing factor in Bank of America subsidiary Merrill Lynch, which the government said failed to submit some required SARs. Bank of America agreed this month to pay $7.5 million to the SEC to resolve the allegations.

Duplicative operations compound inefficiency, Glass said, adding that banks often run separate AML and fraud departments that generate overlapping alerts on similar matters — a redundancy AI can help eliminate.

“Now, when taking that information and analyzing it with AI, I’ve seen banks reduce their alert population by 15% to 20% just [by limiting] the duplication,” Glass said.

AI makes processes more manageable

TD Bank increasingly is using AI as it restructures its AML processes after a 2024 fine of $1.3 billion from FinCEN and a $1.8 billion fine from the U.S. Department of Justice for failing to comply with AML laws.

AML remediation is the bank’s No. 1 priority, a sentiment the Toronto-based bank reaffirmed to FinAi News today.

“Our AML program is now running on a new transaction-monitoring system that has embedded machine learning and AI enhancements,” Leo Salom, chief executive at TD Bank U.S., said May 28 on an earnings call.

“While there’s more work to do in the coming months, I’m encouraged by the progress being made.”

Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.