top of page

When is it Appropriate to Use AI in AML

Updated: Apr 7

On Tuesday, February 27, 2024, the ACAMS New York Chapter hosted an event titled "When is it Appropriate to Use AI in AML?” sponsored by HAWK:AI and in conjunction with the Fordham Gabelli School of Business, The Center for Professional Accounting Practices at Fordham University. The discussion was moderated by Nishi K. Gupta (Counsel, McDermott Will & Emery) and included the following panelists (pictured from left to right, with Nishi last): Mehul Palan (Global Head of AML & Brand Risk Management Analytics & Detection, PayPal); David Choi (Partner, Oliver Wyman); Chris Caruana (Strategy, Hawk AI); and Yariv Ten-Ami (Global Co-Head of Financial Crime Compliance Transaction and Market Surveillance Strategy, Goldman Sachs).

The conversation began by describing what is meant by the term AI and the panel spent some time discussing machine learning as a form of predictive AI and large language models as a form of generative AI. The panel then outlined how predictive AI is designed to predict an outcome, while generative AI is designed to create new content. Both techniques require the use of very large data sets where the data must be normalized and optimized to aid in successful outcomes and successful implementation.

The intention for using either forms of AI for AML processes is not to eliminate human involvement, but rather to leverage human input more efficiently. The panelists gave additional background on the three stages of AI implementation, which are: preparation (cleaning data); execution (training models); and deployment (sustaining performance).

The panelists then went through specific use cases for how AI can be leveraged for the various components of an AML program. These examples included obtaining KYC information, improving transaction monitoring, performing thematic SAR analyses, and establishing ongoing monitoring KPIs. When implementing AI for these or any other purposes, however, panelists stressed the importance of effectively communicating the use of AI to internal and external stakeholders so that there is clear understanding regarding the decision and benefits of using the new technology.

The panelists then took questions from the audience to close out the event.



bottom of page