The IIF surveyed 59 institutions (54 banks and 5 insurers) on their exploration and adoption of Machine Learning techniques in Anti-Money Laundering. While the detailed version of our resultant report is limited in its distribution to the regulatory community and those 59 firms, a short-form summary report has also been prepared for public distribution.
This study covers the particular purposes of application in the AML space, as well as which types of specific techniques are in scope, firms' maturity in adopting, benefits, challenges and model governance. Our findings indicated that the application of machine learning techniques in AML is spreading quickly across the industry, driven by a dedication to build a stronger and more effective defense system against illicit activity. Significantly, none of the 59 surveyed firms were pursuing machine learning as a means to reduce staff, but rather to gain greater and faster insights that can be made available for their trained AML analysts.
Machine learning techniques also hold great promise in addressing some of the challenges financial institutions are grappling with, and can increase the efficiency of the existing AML framework, already helping to drive a reduction in false positive rates and better transaction monitoring results as just one example. We fully expect this trend to continue, with more firms reviewing their processes and launching new projects. We also identify that the improved performance in detection and risk management can be further increased with enhancements in information sharing and feedback loops between public authorities and financial institutions.'