There are huge arguments for arguing that you and/or your business will never be
caught up in any money laundering by being falsely accused of fraudulent
transactions. Well, there’s a reason they say ‘Never say never’.
With clever digital systems that grow and develop almost as we (don’t) watch and
money laundering on the increase, there is some kind of likelihood that just about
any financial transaction online may feature in an anti-money laundering (AML)
search. Even if it’s only a false positive alarm.
All of this can be reduced by using AI in AML systems.
After reading this article, you’ll have a broader understanding of the role of AI in
After reading this article, you’ll have a broader understanding of the role of AI in AML, particularly in reducing the number of false positives.
What are anti-money laundering detection schemes?
Detecting money laundering is about finding any fraud that is committed that is directly related to money laundering. These are schemes or fraudulent transactions where money is moved quickly across borders without being recorded.
The fraud can be committed by infiltrating the IT systems in a bank, or other financial institution, or manipulating crypto currency and other online transactions.
To work against these forms of money laundering, anti-money laundering (AML) detection systems are put in place. In some countries, having AML detection schemes in place is compulsory for any financially-related institution or business. They are software tools that detect the types of fraud that constitutes money laundering.
What are the implications of false positives that are recorded for AML?
On the very simplest level, recording false positives on AML detection schemes means that a transaction is reflected as being fraudulent when it isn’t. The question is: How can false positives be recorded? Is this a reflection of the efficiency of the detection schemes?
False positives are opposed to false negatives, which are fraudulent transactions that are not detected by the AML schemes. Both of these are, clearly, a problem for any financial institution.
The problem with false positives is that they indicate some form of problem with the AML scheme. They cannot be as efficient as necessary, if they can record false positives. The result of this will mean that a lot of admin needs to be done to trace the transaction and deal with it, which takes time and money.
Another problem is that, if a good customer has a transaction (or even more) indicated as being fraudulent, the institution will lose their trust. False positives can lead to further losses for the business.
How is AI used in AML detection?
Artificial intelligence is very valuable in AML, because the machine learning models can look over huge amounts of data and detect any patterns of activity. These are based on algorithms that have been developed and programmed into the learning machines.
If any transactions could indicate money laundering, then the AI will flag them as fraudulent. As digital money laundering practices develop, the AI learns to recognize the new patterns and adjust to be able to detect them.
The basis of AI is that it can learn and develop its own systems as it goes along. If new elements are introduced into an AI system, then it can learn to combine what it already has with the new.
This means that you won’t have to spend time and money redoing all your AML detection systems when you adopt AI. Machine learning will incorporate your old systems with its own and create a hybrid system for your business.
Using AI to reduce false positive rates in money laundering detection systems
Because AI is so widely used and also because it can learn as it goes along, it can also be taught to read a wider range of data than other AML detection systems can. The algorithms can also read unstructured data, such as social media and news articles.
This means that AI can look more broadly for possibly fraudulent transactions more broadly, more quickly and generally more accurately. Apart from analyzing the data from actual financial transactions, AI can look for patterns in communication, or evidence of other contact, between parties who are considered to be suspicious.
Another way that AI can detect hidden relationships or connections between possibly suspicious people or entities, is by using graph algorithms or by analyzing the network. This process will sift through a network of transactions, communication and possible connections that is complex and even confusing. As the AI learns, it will become more and more useful as an AML system and reduce false positives.
What AI can also learn to do is to identify and then analyze illegal or ‘below board’ transactions and interactions, which can also help to reduce false positives. On the flip side, AI systems will also learn to identify legitimate transactions, which won’t then be flagged as possibly fraudulent. This will also reduce the number of false positives.
Leave a Reply