PRACTICAL SOLUTIONS
It is accepted that an AML statutory and regulatory framework needs to be continually updated to detect criminals and counter terrorism in the 21st century . 2 The current regime faces multiple persistent obstacles , including high false-positive rates , increasing regulatory hurdles and evolving criminal tactics .
Traditional rule-based AML systems , which are still prevalent , generate false-positive rates as high as 90-95 %, resulting in an overwhelming workload for compliance officers and increased operational costs for financial institutions ( FIs ). 3 Professional money launderers employed by global threat organizations commit crimes in ways that appear legitimate and fail to raise red flags through conventional detection processes . This underscores the urgent need for more accurate and affordable solutions .
Potential benefits
The integration of ChatGPT into AML systems presents an opportunity to address these challenges by leveraging its advanced natural language processing capabilities . Through sophisticated pattern recognition and anomaly detection , ChatGPT may reduce false-positive rates . Moreover , the model ’ s ability to learn and adapt from large data sets may allow for prompt identification of emerging money laundering schemes , thereby enhancing the system ’ s responsiveness to evolving threats .
In addition to possibly improving AML detection capabilities , ChatGPT may be used to streamline AML compliance processes . Utilizing automation as an intelligence-driven “ investigator-centered approach ” may increase detection . 4 For instance , if the AI-driven model could automatically analyze and summarize complex regulatory requirements , it may enable compliance officers to stay informed and implement appropriate controls more efficiently .
The deployment of ChatGPT in suspicious activity reports ( SARs ) may improve information sharing among FIs , regulators and law enforcement ( LE ) agencies , enabling more successful collaboration and a stronger collective response to money laundering and other financial crimes . In the U . S ., for example , “ 95 percent of investigations submitted in response to information-sharing requests by the Financial Crimes Enforcement Center ( FinCEN ) yield positive results . Without such information sharing or tangible leads of some kind , less than 2 percent of alerts achieve productive results .” 5
While suspicious activity detection may be a primary target for the use of ChatGPT , there are other uses that could enhance the effectiveness of AML departments . Documentation gathering is a time-consuming process for case reviews conducted by analysts and could likely be handled by an AI component , allowing for less time spent on monotonous “ check the box ” activities and more
ChatGPT may be used to streamline AML compliance processes
concentration on making consequential determinations . At the back end of the process , it appears plausible that ChatGPT may be able to complete currency transaction reports and SAR filings , even going as far as writing SAR narratives utilizing the facts on hand . This would allow analysts and investigators to utilize their time to make a determination and check for accuracy , making the entire review process faster .
Many Office of Foreign Assets Control ( OFAC ) and watch list-related screening tools use logic that flags variations or partial matches of names that clearly do not match the applicable entity . ChatGPT could potentially be leveraged to clear these hits , greatly reducing “ noise ” in alerts and allowing focus on only those that most closely align with the flagging entity .
In addition , with many institutions adopting some sort of model to detect potential suspicious activity , ChatGPT may be able to assist with model tuning for greater efficiency , as well as above- and below-the-line testing . Undertaking this process manually can be laborious , but by condensing the time and potentially enhancing accuracy , the overall performance of these tasks may be improved .
In summary , ChatGPT may help to reduce the risk of financial crimes . Money laundering and terrorist financing pose a significant threat to the global financial system . 6 ChatGPT may aid in detecting suspicious activities before they can be conducted , thus protecting FIs and their customers from the negative consequences of financial crimes , such as reputational damage , loss of funds and legal liability . Indeed , it is currently estimated that a regulatory fine is a top-five loss for any bank . 7
44 acamstoday . org