COVID-19 is accelerating the AML and CTF pivot toward agile AI systems

The COVID-19 crisis has created new challenges for the compliance department when it comes to money laundering and terrorist financing. Financial institutions are fine-tuning their policies to respond to the warnings issued by regulators, and at the same time, updating risk-mitigation measures connected to the crisis. Understanding new risks and adapting operational responses are the fundamental problems that compliance managers currently face. The crisis has, however, created great potential for managers to move away from a tick-box approach to execute the AML compliance requirements. It marks a new beginning in which the compliance team has the opportunity to take the reins of the financial crime fighting battle.

We compliance professionals are well aware of the fact that around 1 % of criminal proceeds from money laundering activity is confiscated. This is inadequate and can be demotivating. From my personal experience, I remember how unsatisfying it was to have a high volume of work and high backlogs of cases and, at the same time, a low level of achievements and not seeing results. There are several hurdles that need to be overcome which are currently contributing to this low 1% number:

• Data aggregation from various sources
• Fragmented systems
• A high number of false positives
• A high volume of backlog / Workload problems
• Lack of Human Resources
• No feedback from FIUs to the financial institutions

Compliance teams must all too often use significant resources to collect data from different systems instead of spending their time on investigative work. At the same time, authorities during this crisis suggest that companies should not change or switch off transaction monitoring thresholds or sanction screening systems for reducing the number of alerts. This leaves compliance teams in a tough spot.

With all of these hurdles to overcome, how is it possible to tackle operational challenges, and at the same time, minimize the risk of money laundering and terrorist financing? Here are three areas that can lead your organization in the right direction:

1. Improve your controls

The necessity of improving controls through better segmentation of alerts, using the KYC/ Onboarding data, and other relevant data from 3^rd party providers can help reduce false positives and create a much higher level of effectiveness. This data aggregation and segmentation must be automated so as to save compliance teams time. Furthermore, it is important to have flexible rule-based monitoring in place that can be quickly adapted to, for example, new observations made during the crisis. As behavior changes, controls and rules must also be adapted based on market expertise, knowledge and trends.

2. Machine Learning

Machine Learning can provide additional insights into customer activity while significantly reducing operational costs and increasing the effectiveness of investigations. It offers a perfect opportunity to merge both knowledge based rule controls with artificial intelligence. Machine learning models can complement the AML strategy and provide a benchmark to help support human decision-making. Certain types of data anomalies and suspicious transactions can be detected using supervised machine learning systems. Unsupervised machine learning algorithms can also be used to find and focus on the riskiest transactions that have not been detected by looking at customer profiles or assessing transaction behavior versus peers.

Machine learning algorithms, such as unsupervised anomaly detection, can help financial institutions analyze large volumes of data and, at the same time, help to identify previously unobserved patterns. Such models can be trained for different risk scenarios, and implementing these models will create a double-safety net for the organization.

Using machine learning can enhance the detection of deviations from normal behavior. Looking for anomalies is essential in the process of fighting money laundering and terrorist financing. The key objective is to develop a model that is simple to implement and sophisticated enough to protect the financial institution against money laundering and terrorist financing.

3. Workload Optimization

As the practices of the banks have shown, it is possible to prioritize alerts and route them to the appropriate team. Manually routing the alerts can consume a significant amount of resources and result in false positives ending up with the 2^nd line CAMS specialist. Over time, through implementing machine learning, predictions can be made as to when an alert is sent to the 1^st line or when it can go directly to the 2^nd line, thus skipping the first step and saving time and money. This will result in more investigations of high-risk transactions rather than low-risk suspicious alerts. Optimizing workload in this manner can significantly improve the investigation efficiencies.

Closing thoughts

A transition toward AI systems can only happen with the essential support from regulators. In order to address the hurdles mentioned above, I believe the way forward must include the widespread adoption of today’s technology. Using data harmonization and aggregation platforms, machine learning, fuzzy logic, and advanced analytics to assess risk across products, geographies, and processes will prove to be the most efficient way to enhance the fight against financial crime. It is time to bring that 1% number up and stop the criminals in their tracks.

How do you think we compliance professionals can improve on the 1% number?

---

---

You may also like