Putting the ML into AML

The FCA and Bank of England have recently published the findings of their first ever survey of the use of machine learning (ML) in financial services firms. The conclusions from the survey are that there has been significant uptake and deployment of ML in financial services firms, particularly in larger insurers and banks.

The survey suggests that ML is most frequently used in back-office and other support functions, because these are the functions performed by all financial services firms and also these tasks often lend themselves to automation. AML and fraud detection are given as examples of back-office functions which have seen high levels of deployment of ML.

In this context, it is interesting that in a speech on Wednesday, Megan Butler, Executive Director of Supervision (Investment, Wholesale and Specialists) at the FCA, issued a call to arms to the financial services industry to make greater use of AI and ML to combat money laundering, fraud and other financial crime. The FCA is expressly technology neutral, but there has been a trend in their recent publications of giving prominent recognition to the opportunities offered by technology (and particularly AI and ML).

Using AI and ML in a financial crime context makes sense, given that the detection and prevention of money laundering and financial crime require the connection of large data sets and associated pattern recognition. Encouragement from the FCA for firms to experiment in this area, and an acknowledgment that not every experiment will be successful, will only drive more firms to explore the potentials of ML.

It is perfectly possible that use of ML in an AML context will become the industry standard in the next 5-10 years, especially given warm noises coming from regulators. However, this does not mean that ML can be deployed without careful thought and robust governance. Firms will increasingly need to shift their thinking from seeking to identify tools which can assist in their compliance, to making sure they've got sufficiently robust structures and procedures in place to oversee and monitor the effectiveness of the tools they've deployed.

In that context, the demise of the compliance professional may have been mourned prematurely, but the role is likely to shift focus in the coming years.