On the Potentialities and Limitations of Autonomous Systems in Money Laundering Control

Abstract

This paper analyses the potential gains and eventual difficulties using autonomous systems – such as artificial intelligence (AI) mechanisms – to prevent, detect and investigate money laundering. As it is well-known, new technologies have been applied in the most varied social contexts, being no different in the case of the FIUs, especially when receiving and processing reports of suspicious activities from obligated entities. However, in addition to the already identified difficulties imposed by new technologies, the specific scope of money laundering presents particular challenges. Potential guidelines are proposed for a better interaction between AI and money laundering prosecution. For that, is is initially analysed what is effectively meant by AI and autonomous systems and how they are effectively used in this scope. Subsequently, some of the difficulties encountered in this context are demonstrated, ranging from insufficiency, low quality and inaccuracy of data that feed the systems, to the difficulties in understanding, explaining and allowing the refutation of the conclusions reached by them. From this analysis and through a deductive methodology, possible solutions are proposed that allow a better and more efficient interaction between humans and autonomous systems in the field of money laundering and its prosecution.