Share This Page


Share This Page


Today, as the world rapidly embraces digitization, the emphasis on fighting financial crime, money laundering and terrorist financing using cutting edge technology have also increased. Due to the global phenomenon of digitalization, the amount of ‘wire’ activity is growing significantly, increasing the pressure on banks and financial institutions to monitor and detect suspicious activity prudently and making it harder for them. Most banks still use using archaic tools; they have an immature services ecosystem and depend on manual rules creation with no self-learning capabilities.

At the same time, in the recent past, we have witnessed that regulators are imposing hefty fines on banks and demanding a higher level of scrutiny on banks and financial institutions in the transaction monitoring space with smarter approaches. To quote, one of such real time incident from The Sydney Morning Herald- “Australian regulator Austrac alleges Common Wealth Bank (CBA) in serious breach of money laundering act. Austrac has launched civil proceedings in the federal court alleging that CBA failed to comply with the AML and terrorist financing laws on more than 50,000 occasion.”

After spending 25 years in the BFS industry, what I have realized is that with growing datasets, disparate transaction data systems, integration issues with monitoring systems and changing regulatory norms, this is leading to newer rule creation every time, which is humanly impossible. Hence, AI/ML is coming into play in this space, as now, banks need to move beyond the rules engine and adopt dynamically adaptive predictive models or workflow engines, which would enable them towards real-time, transaction-based KYC anomaly detection and highly refined self-learning models focused on anti-money laundering SAR (suspicious activity report) detection. The significance of analytics in combating AI/ML have recently been realized by industry experts. Of late, regulators recognized that rules alone are not an effective way of detection and are thereby pressuring banks to include more advanced analytics.

Let me also highlight some of the problems/ challenges financial institutions face:

  • Setting up the correct threshold levels and parameters - When thresholds are too low, the system will populate a high number of alerts that require analysis. On the other hand, if thresholds are too high, the amount of alerts will decrease. However, the company may not detect all suspicious activities and fail to meet regulatory requirements, which in turn risks both reputation and exposure to fines.
  • Identifying ‘false positives’ quickly and accurately- Analysis of the alerts may be time consuming, but must be completed with a sufficient level of scrutiny to ensure compliance with existing governance processes. False positives, which are likely to be the biggest challenge, should be identified at the right time and removed as quickly as possible.
  • Complying with global and regional laws and regulations
  • Accurate and timely reporting
  • Streamlining operations to minimize costs

The only answer to all the aforesaid issue is apply AI/ML in transaction monitoring. Banks/ FIs need to have a robust decision support system by leveraging a ML-powered predictive analytics platform which uses a self-learning mechanism on the predictive models to continuously evolve with new data points and user analysis. The platform will help in using high accuracy decision support using model prediction and confidence scores. It must also have detailed audit features for recording of model output to trace the rules creation. This will help banks improve continuously in terms of prediction accuracy and drive operational efficiency. In parallel, this will help banks move from static rule-based applications to a dynamically adaptive system, thus driving significant operational efficiency in its processes.

In order to identify spikes in value or the volume of transactions, monitoring high risk jurisdictions, identifying rapid movement of funds, screening against sanctioned individuals and politically exposed persons (PEPs), and monitoring enlisted terrorist organizations, we are currently witnessing increasing examples of ML in many areas of technology. Banks/ FIs should grasp this opportunity and use it for repetitive analysis. To fulfill this wish of banks, IT solution companies must try to build a robust predictive analytics platform using a leading ML platform which will help banks move from a static rule-based model to dynamically adaptive predictive models. To combat with all the hitches mentioned, I feel the platform must have the following features:

  • Dynamic Model - create accurate models using assets of mathematical model for dynamic workflow creation
  • Self-Learn – in-built capability to keep models updated through self-learning technology with the ability to identify only those transactions which are genuine risks, to be reported
  • Audit – Often, ML models are treated as a black box. Real-time audit capabilities will give an edge to business as regulators want every alert to be audited
  • Prophecy with confidence scores – High accuracy with confidence scores provide business comfort to take decisions based on model predictions
  • Deployment- Faster time to market by deploying models in an internal production environment

To conclude, I would like to state that by replacing legacy systems with a leading ML/ AI platform, banks can substantially reduce operational costs, effectively tackle highly unbalanced data, improve alert prediction over time, increase reporting accuracy and improve compliance.


About the Author

Subhasis Bandyopadhyay
General Manager, BFSI

Subhasis Bandyopadhyay heads the BFSI practice at Mindtree. He is responsible for offering leadership and direction for BFSI solutions, domain consulting, alliance management and domain competence building.

Other Stories by the Author

Let's Talk About Your Needs

Thank you for your submission. We'll be in touch.