Over the last years, regulators have gained extensive powers to impose fines and sanctions: US regulators have taken the lead, while European supervisors play catch-up. AML fines in Europe and the UK totalled $214M from 2014 to 2017, with those in the US at $1.96Bn. Fast forward to last year and, during the first three quarters of 2018, fines in the UK and Europe reached $918M, compared with just over $1Bn in US penalties. Hence, FIs have been starting to deploy the new generation of smart weapons rolling out machine learning innovation across different stages of the Financial Crime detection process (see figures below).
Via the replacement of rule-based models with data-driven ones, the suspicious activity detection process can be made more objective, time-efficient and cheaper. Furthermore, Big Data analysis techniques can help to spot hidden patterns (e.g. particular cases of "smurfing") and processing external unstructured information (e.g. adverse media detection).
Needless to say what the final aim of all this digitalization process is: the reduction of false positives. 99% of the FIs rule-based-generated alerts, in fact, the result being false alarms and, hence, are not escalated into SARs; involving machine learning has proved to be a winning strategy reducing to less than 70% the rates of false positives detection.
Why, though, if the upsides are so many, hasn't AI been actively implemented by the majority of the FIs yet?
Thanks to our hands-on and industry-specific expertise, at Alpha Reply we can offer help throughout any of the stages of the process by taking complete ownership of the whole project, or via offering team reinforcement capabilities.