6. Oct

Is automated technology effective in combating money laundering?

With the growth of the digital world, verifying identities and effective transaction monitoring have become increasingly difficult despite the onslaught of software which promises you the golden ticket; Anti Money Laundering (AML) safety, security and most importantly, no breaches. Previously AML processes within organisations have been complex and costly. Millions of transactions are carried out every day around the world, and it isn't easy to track them successfully. Tracking and verifying information is a very manual, labour-intensive job. The solution was obvious, software.


Google “AML software packages” and be prepared to spend the new few hours trawling through pages and pages of offerings. How in a matter of a year has the need for AML automation grown to this level? One word: Trends.

A report found financial institutions spent US$181 billion on financial crime compliance worldwide in a year with European firms spending three to four times more than north America. When it comes to transaction monitoring, financial institutions across the board are generating an average of 90% false positives.

Sphere heading the drive towards AML technology solutions are the fines imposed against banks and businesses. With a fine comes three things every institution and business aims to avoid (Even more if you are a listed company):

  • Reputational Damage
  • Loss of Customers
  • Loss of Revenue
  • Impact on share price (if a listed a company)

Let’s take a look at the technology being implemented globally to reduce AML/ Counter Terrorism Funding (CTF) breaches.

Machine math 2120x670.jpg



This is software that can be programmed to do more basic tasks across applications and systems. The aim is to reduce the need for humans to complete repetitive tasks. It can be used in the Know-Your-Client (KYC) process to gather data from various sources and comply the information in one location almost like a copy and paste type function. It can be used to gather information to complete a comprehensive credit check in a matter of seconds.


AI aims to have computers performing tasks the way a human would. It seeks to replicate human thinking and logic to form a prediction.  AI creates a program for a specific task and allows it to improve on its own. AI systems are expected to come up with a solution when faced with atypical situations verses grinding to a halt. To break this down even further Microsoft Word cannot improve on its own, but facial recognition software can get better at recognizing faces the longer it runs. AI examine large data sets and create patterns which they use to make predications. The more data it “absorbs” the better the predications.


Machine learning allows computers to learn from their past mistakes without being explicitly programmed to do so. The initial programme applied to the system includes an algorithm which allows for the learning to happen. This improves the accuracy of the information it delivers. It is a subset of AI. ML is the most impactful when used for transaction monitoring as the software rationalised and reviews the information presented to it but also relies on “muscle memory” from previous experiences to make a determination on whether a transaction is out of character for the client or if an there is an element of fraud attached to the transaction.

ML and AI are seen to be the most beneficial in the AML/CTF processes as they:

  • Reducing the rates of false positives through semantic and statistical analysis of readily available customer data to correctly identify high risk entities.
  • Detecting data points within transactions and customer behaviour with a high potential of fraud.
  • Create Benchmarks for client’s behavioural patterns based on the transaction history and real time behaviour.
  • Tracking daily changes in customer behaviour and monitoring those who have diverted from their usual patterns.
  • Has the ability to identify gaps in the customer information received based on changes to regulations or international requirements.
  • Determining whether a given deviation in customer behaviour is acceptable or if it needs to be investigated further.
  • Automate the updating of a client’s risk profile throughout the client’s life cycle based on all of the above noted points.
  • Decodes and detangles vast amount of data to easily uncover the Ultimate Beneficial owners of complex structures.
  • The ability to detect patterns in a large amount of text enables it to comprehend the consistently evolving regulatory landscape.



Before making the investment in automated system here are a few things to consider before picking an option;

No two organisations are the same in terms of client base, geographical location and product offering. You need to ensure the package you chose is right for your business and can account for any nuances.

Automation will speed up onboarding times and reduce human involvement in transaction monitoring, but you still need to know your customers; Who is on the sanction list? What classifies a Political Exposed Person (PEP)? What level of shareholding determines an Ultimate Beneficial Owner (UBO)? Are there indirect sources of control? Are you comfortable with the organisational chart? While automated technology helps identifies abnormalities, you still need to know all the “low risk, normal” clients in your client base.

You need to ensure the information being classified as low risk aligns with your appetite for risk and your own internalisation of risk exposure you are happy to bear when onboarding new clients.

Amanda O'Donnell Compliance and Corporate Secretarial Manager