FIs must prioritize strong governance and robust data when implementing AI systems to improve anti-money laundering processes.

FIs should “start out with strong governance and build the foundation of the program from there and then also take a look at their data and data integrity,” Jon Glass, a partner in the financial crimes advisory practice at consultancy SolomonEdwards, told FinAi News.

“Because if the data is all messed up or convoluted, that’s going to make AI less reliable,”

ALSO READ: Implementing AI can relieve bottlenecks amid growing AML complexity

AML AI requires comprehensive, up-to-date data for training and efficient operation, Glass said.

Although AI looks at past trends, incidents and datapoints to forecast risk and prioritize workflow, ensuring the data is collected and properly categorized, which is critical, requires perseverance from leadership, Glass said.

“AI cannot fix poor governance. It cannot fix weak data or inconsistent processes,” Glass said.

Explainable AI

While FIs must comply with expanded regulations and bureaucracy before implementing AI, bad actors have no such restraints, Tracy Moore, director of thought leadership and regulatory affairs at fintech Fenergo, told FinAi News.

“The technology is great for the banks as long as they implement it and use it and get the foundation and the infrastructure ready for the advancements in technology,” Moore said. “But I think the fraudsters are faster.”

Straightforward AML processes and transparent results can help FIs clear regulatory hurdles, Glass said.

“Strong governance provides the explainability, the oversight and the data discipline that regulators expect before AI can be deployed broadly,” he said.

Explainability is a key factor in driving AI implementation because the financial industry carries a heavy regulatory burden, Shanon Mclachlan, chief operating officer at tech provider Jack Henry, told FinAi News.

“Any AI solution deployed cannot just be effective — its decisions must be entirely explainable and auditable under strict model risk management frameworks that govern, validate and monitor model behavior,” he said.

Helping hand

While some new regulations place stumbling blocks in front of AI implementation, others provide frameworks for its use.

“There has been a gradual increase in regulatory expectations across multiple dimensions and obviously AI is one of them,” Madhu Nadig, co-founder and chief technology officer at tech firm Flagright, told FinAi News.

“But increasingly we’re seeing that regulators themselves are proactively coming up with frameworks that allow FIs to actually leverage AI because regulators do understand that the current systems are kind of broken.

“I wouldn’t say there is just more regulatory stress on organizations,” Nadig said. “There are also frameworks that allow organizations to meet their requirements in a more forward-looking way.”

He said regulatory bodies at the forefront of providing these frameworks include:

• The U.S. Department of the Treasury, through its Financial Crimes Enforcement Network;
• The Inland Revenue Authority of Singapore, through its Major Exporter Scheme; and
• The United Kingdom’s Financial Conduct Authority.

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