One of the biggest challenges for banks, payments firms, fintechs, and wealth and asset managers is to correctly interpret regulatory guidance and turn it into anti-money laundering (AML) strategies within their organisations. But forward-thinking regulators are now collaborating with the financial services industry to close this gap. Different participants within the ecosystem bring specific expertise as to what regulatory mandates look like when translated into production within financial institutions technology environments.
For many financial institutions (FIs), they feel stuck between a rock and hard place when it comes to delivering on the aims of the regulators to better combat financial crime. FIs know that they need to deploy new AI innovations to combat emerging threats and deliver the automations necessary to conduct AML operations at the scale of modern transaction volumes. But the complexity of developing these models has led some FIs to test ‘off the self’ generic models that cannot be tuned to their specific risk appetite, and often do not meet regulatory requirements for explainability. So called, ‘black box AI’ may initially appear to deliver positive results, but will likely result in a high number of false negatives and ultimately, non-compliance.
A compliance first approach to Transaction Monitoring typologies:
Napier AI prides itself on a compliance-first approach. What that means is that rather than offering opaque/closed AI which may generate alerts, but cannot be well understood by AML analysts, we have developed a library of typologies within our Transaction Monitoring solution. These typologies are built on the work conducted with the FCA, and mapped to our data model, which allows financial institutions to easily select the typologies most relevant to their business and tune these in line with their risk appetite.
How to detect AML typologies
1. Rules that looks at single or aggregated transactions against specific values/frequency/thresholds etc., such as high value or volume of transactions over a set period of time.
2. AI behavioural detection that looks at transactional behaviour indicators in relation to historic activity of the entity and the peer group/segment of the entity, such as excessive withdrawals or deposits.
3. AI network analytics that look at the combination of transactions and behaviours across a network of connected entities, such as identifying circular payments as criminals move funds between their own accounts and those of close relations in an attempt to lose the financial institutions tracking them.
The 100+ typologies in the Napier AI library are defined including a description of how they commonly occur, the key indicators, and themes that correspond to similar typologies. Themes provide a useful link to non-identical typologies that share features, for example the use of cash, or cross-border transactions.
Typologies continue to proliferate, and criminals are continuously employing new technologies to evade detection. Through our work developing a global typologies library and mapping it to the Napier AI data model and our detection capabilities, we have been able to develop a streamlined strategy for tackling the most common typologies presenting risk by segment of financial services. And FIs can rest assured that their financial crime compliance approach is tackling typologies in line with the recommendations from global regulators.
Napier AI’s transaction monitoring was recently named as a Technology Standout in the Celent AML Transaction Monitoring Report. Read the full report analysing 16 vendors in the AML space
