To Combat Money Laundering, AI Tools Come Off the Shelf and In the Cloud

The battle against financial crime and money laundering is as much technological as it is a matter of law enforcement and regulatory compliance. As artificial intelligence has proved to be potent in detecting and deterring criminal activity, the technology is more and more accessible to banks and others on the front lines via software-as-a-service (SaaS) and cloud computing.

Enterprising vendors like Fenergo and NICE Actimize grew from the start-up stage to established players in anti-money laundering (AML), Know Your Customer (KYC) and adjacent compliance and surveillance functions. Joining the fray this year, and representing further mobilization of advanced technological tools toward these ends, is Google Cloud. Its AML AI product uses risk scoring to improve upon rules-based approaches to identifying suspicious transactions, according to a June announcement.

“Google Cloud customer HSBC found that they can now detect two to four times more true positive risk, enhancing their ability to identify and prevent money laundering activities,” the Big Tech leader said, while speeding up investigations and reducing alert volumes by more than 60%.

"As threats become more sophisticated globally, and the challenges in fighting money laundering become increasingly complex, we believe in the combination of AI and decision science as the best strategy to detect suspicious activity with more accuracy and efficiency," said Rafael Cavalcanti, senior vice president Data & Analytics at Bradesco, another Google Coud AML AI user.

Silent Eight’s Martin Markiewicz: ‘Organizations must continually adapt.”

On November 1, integrated surveillance provider SteelEye launched Compliance CoPilot, “a tool that uses the capabilities of advanced Large Language Models (LLMs) to greatly accelerate the Communication Surveillance Alert Review process.” Risk-based analysis and breach identification over email, chat and other methods occur at much faster than manual speed, and in multiple languages. “In all instances, the rationale for its conclusions is given to provide critical evidence to compliance users,” SteelEye said.

Firm Size and Complexity

How effective is the increasingly accessible proliferation and variety of AI anti-fraud tools?

“The viability of an off-the-shelf SaaS product for AML and fraud detection varies based on the organization's size, resources, and the complexity of the threats they face," says Martin Markiewicz, CEO of anti-financial crime technology company Silent Eight. They can be most cost-effective for “smaller institutions like Tier 3 FIs and non-FI corporates,” while “larger organizations with substantial budgets may require a more comprehensive approach that integrates various technologies, subject matter expertise, and extensive data analysis to stay ahead of evolving threats.

“Regardless of the chosen approach, it's crucial to recognize that combating financial crime is an ongoing battle, and organizations must continuously adapt to the ever-changing landscape of money laundering and fraud,” Markiewicz added.

Ricardo Amper, founder and CEO of identity solutions company Incode Technologies, suggests adding that layer to the protections of AML software. “You have to consider Politically Exposed Persons (PEP) watchlists, banking laws, the USA Patriot Act and other mandates that keep AML operations current,” Amper says. “In addition to verifying identities and ensuring AML regulations are met, SaaS products need to deliver a seamless customer experience, integrate with other enterprise technologies, lower costs and scale for operational complexity."

Establishing an Ecosystem

“As with other tools that have come before, the process of detecting AML and fraud activity is constantly evolving,” notes Peter Kwan, senior director in the Forensic Technology Services practice, Alvarez & Marsal Disputes and Investigations.

“At this stage, I am not sure if any tool will function as a silver-bullet solution that fends off AML and fraud activity for good,” Kwan continues. “However, I do think that more advanced detection approaches, from machine learning (ML) models and cloud-based, to always-on architectures, are necessary to evolve our capabilities.

Alvarez & Marsal’s Peter Kwan: Value in information-sharing.

“There is also value in having a more centralized, yet anonymous, information-sharing mechanism to enrich AML and fraud activity detection – such as an approach that is deployed over multiple banks, either in the same region that target similar customers, or that are susceptible to the same AML and fraud typologies."

“Banks need a comprehensive ecosystem of compliance tools to maintain a robust AML program, particularly when operating on a global scale,” says Fenergo head of financial crime Rory Doyle. “To illustrate, consider two French banks that maintain correspondent banking relationships with a Cameroonian bank. One of these banks might possess branches in Cameroon due to the historical ties between Cameroon and France, potentially leading to a higher risk tolerance in their Cameroonian operations compared to the other bank, which lacks a relationship with the Cameroonian regulator."

In that context, generic compliance is not sufficient. The system must be geared to the risk and risk tolerance. “A customer may be a PEP (politically exposed) in Cameroon but not in France, or a prospective client might be PEP status in Ireland but not in the U.S.

"To highlight how intricate it can get, take JPMorgan, a bank in New York which has a branch in Ireland. Such a global presence necessitates the establishment of a comprehensive compliance ecosystem to effectively navigate and address these intricacies and challenges."

Experts and Systems

Matthew White, co-chair of the Financial Services Cybersecurity and Data Privacy team at the Baker Donelson law firm in Memphis, Tennessee, says that financial institutions vary in their reliance on consultants to develop their AML compliance programs. Their choice is often influenced by the size and resources of the institution, as well as the state of the institution’s program.

New technology tools “will likely alter, rather than replace, the work of firms specializing in AML compliance,” White says. “I would expect those firms to begin utilizing these AI-based solutions as a component of their compliance offerings.”

Baker Donelson’s Matthew White: AI as a component.

He sees an analogy to document management consulting: “With the increased availability of AI tools, many of these consultants are using AI-based tools to assist in the document mapping process, and then will use that information in creating broader document management policies and procedures. We may see a similar dynamic here, where consultants will use/recommend tools like Google’s to assist in the risk detection process, and then will continue to provide additional support to financial institutions in developing and implementing the overall AML compliance program and ensuring that it is customized to the institution’s risk assessment.”

Still, more help will be needed as the challenges continue to multiply. AI and ML tools can be invaluable in sifting through mountains of customer and transaction data, but there are still gaps in their adoption. They ultimately must be easy to implement and provide accurate information that lightens the load of compliance and risk personnel.

Cost Efficiency

SteelEye’s 2023 Compliance Health Check Report found that compliance costs increased at 76% of firms in the last year. The company positioned its AI CoPilot to deliver “significant cost savings by boosting analyst efficiency, and the overall scalability of the compliance function.”

Productivity – and the need to overcome people shortages – underlies the sales pitch for the “AI digital workers” of WorkFusion. In a survey it conducted with Celent, 74% of financial institution compliance, operations, risk and IT respondents were unhappy with current staffing levels, and 22% were understaffed.

“For any AML and fraud detection tool to be successful, it needs to cooperatively share the data it gathers with, and learn from, the compliance community,” says Alvarez & Marsal’s Peter Kwan. “Current AML and fraud detection tools can be overly subjective and often operate in isolation. It is increasingly apparent that more efficiency is required, and greater innovation is on its way.

“Cloud-based technology and machine learning algorithms make it easier than ever for data sharing to take place,” Kwan adds, as he underscores the value of cooperation. “The real question is, Will institutions heed the call to the greater good to help further reduce money laundering and fraud, or will they simply isolate compliance to their own organization?”