Combatting financial crime: Machine learning
Exploring the potential of machine learning, and of adopting a joined-up approach to anti-fraud and AML programmes
Is machine learning the future of compliance? That was the question posed to compliance experts gathered in Geneva for Sibos 2016.
Machine learning is a subset of artificial intelligence (AI) which enables machines to detect patterns and make decisions accordingly. While this technology is in its infancy where banking is concerned, compliance is a particularly interesting area of focus. Machine learning is being used by banks for areas ranging from high-end investigations through to KYC at the onboarding stage.
As one expert said, compliance teams currently spend massive amounts of time on false positives. The benefit of machine learning technology is that it can recognise patterns in order to discard false positives and focus on genuine risks.
When it comes to building a business case for machine learning, the panellists noted that key performance indicators include time, efficiency and productivity gains. In order to be attractive, new technology needs to be more effective than existing processes.
Fraud and AML: Time for a joined-up approach?
A separate panel of experts debated whether it is time for banks to rethink how they organise their fraud prevention and anti-money laundering programmes. Most institutions separate their anti-fraud and AML teams, panellists said, even though both disciplines have very similar objectives, face similar challenges and use similar tools and processes.
In light of this overlap a more consolidated approach could provide considerable benefits, including better customer protection and a streamlined process for new customers, panellists said.
Nevertheless, opinions varied about how to achieve this. One panellist said that their joined-up approach involved client education and having more client outreach covering both areas at the same time. Another said that opportunities could be found in the area of investigations case management.
While acknowledging the challenges posed by internal siloes, panellists encouraged banks to think even more broadly. They cited the benefits for banks of collaborating across institutional lines.
One expert pointed out that even if individual banks have strong processes in place, criminals can use relationships with multiple banks to conduct illicit activity while “making things look completely clean within one entity”. The challenge therefore lies in bringing the necessary intelligence together across multiple institutions, as well as between AML and anti-fraud departments internally.
Both panels concluded that while compliance continues to present major challenges for financial institutions around the world, the industry is working to increase the efficiency of the associated challenges. Leveraging machine learning and adopting a more holistic approach to anti-fraud and AML are two ways in which institutions may be able to overcome their current challenges and achieve efficiency gains.
The conversation will continue at Sibos 2017 in Toronto.