How to tackle Anti-Money Laundering challenges through AML optimization and artificial intelligence

co-authored by Sascha Steinbach-Vulin

A constant increase in alerts generated by financial institutions´ Transaction Monitoring Systems (TMS) has significantly elevated the difficulty for Anti-Money Laundering teams to identify relevant AML & CTF-related (Counter Terrorism Financing) transactions. Ongoing dynamics of the regulatory environment of AML laws and requirements have been constraining banks to adapt their search algorithms and thresholds on a regular basis resulting in an unpredictable rise in potential alerts of which – at the same time – an increasing number remain false positives. As a result, financial institutions´ AML expenditures on additional transaction monitoring analysts and investigators have boosted over the past years, leading to substantial overhead costs within most AML departments of major European banks.

Among the three more common challenges that banks are facing within the AML area, we find a high manual effort in transaction monitoring and case management processes, difficulties in maintaining a consistent alert and case handling standard as well as uncertainty regarding alert and case output and prioritization.

Most of the relevant information for case reviews is scattered in various core IT systems leading to a large amount of time spent in finding the data. It is not surprising that many leading banks have inefficient and redundant alert handling processes that increase the time spent for the evaluation of each alert. Moreover, a permanently increasing number of alerts explains the continuous trend of growing AML teams.

Ambiguous operating procedures for alert handling and case management, frequent changes of AML team members and team responsibilities as well as a dynamic regulatory environment, all together lead to confusion among AML teams and to difficulties to maintain consistent standards.

Finally, many elements create uncertainty regarding alert output and prioritization:

• a disproportionately high percentage of false positives regarding overall alert output
• a predominant confusion, and no clear guidelines regarding alert and case prioritization coming from regulatory institutions
• an existing disagreement regarding the definition of relevant thresholds and parameters for alert identification within the industry, create uncertainty regarding alert output and prioritization.

To avoid bloated AML teams and to reduce the risk exposure to fines and penalties from regulators as well as to control the operational risk, banks should be advised to start pilot projects in order to test different solutions and approaches that solve well-known problems.

Regarding high manual process efforts, a detailed analysis of automatable workflows as well as their implementation by utilizing RPA should be taken into consideration. The automation of manual and repeated processes can significantly improve resource capacities and overall work efficiency by an improved ability to review more cases within the same amount of time. During the analysis of a bank’s AML processes, it is essential to identify inefficiencies and redundancies to design and implement an optimized process workflow following industry best practices.

The analysis of existing thresholds and parameters for alert generation is a further area to deep dive. The exploration of information within the existing prioritization approach is key to identify potential gaps. By the utilization of Machine Learning, it is possible to improve alert output and to facilitate alert prioritization.

An AML optimization program aims to improve overall work efficiency within the various levels of banks’ AML & CTF activities by both streamlining and optimizing existing workflows as well as by analyzing and identifying automation and digitization potentials for relevant processes. By doing so, banks will not only reduce penalty risks for regulatory breaches but at the same time promote cost reduction potentials.