For anti-money laundering processes to be efficient and effective through AI and machine learning, data quality must improve radically, writes Livia Benisty, head of AML for payments specialist Banking Circle

Compliance is generally thought of as a necessary burden, not a business benefit capable of vastly improving efficiency. However, applying artificial intelligence (AI) and machine learning (ML) enhances the precision of the rules used in traditional automated AML processes. This cuts down false positives, reduces operational workload and frees up resources to focus on other areas such as customer relationships.

Traditional rules-based processes capture only one element of a transaction, resulting in false positive rates of 97 to 99%. AML that employs the latest AI and ML capabilities, on the other hand, provides a series of indicators that point to something being higher risk. This reduces false positives while still ticking the compliance box and successfully combating financial crime.

Digitalisation has accelerated across financial services in the past year, but regulation and banking processes have struggled to keep up. This has contributed to a rapid increase in money laundering. Each year, to reduce the damage that could be caused by money laundering fines, banks spend an average of US$48 million on know your customer (KYC) and AML processes, with US banks spending more than US$25 billion a year on AML compliance. 

AI is growing

A wide range of financial institutions have begun to introduce AI-based approaches to combat the rise in money laundering. But this isn’t without its challenges. When we spoke to 300 senior decision makers in European banks, we found a widespread belief that AI implementation to date has been far too inconsistent, potentially compromising their business objectives.

Interviewees believe that AI and ML are absolutely essential in the battle against money laundering in the digital future. And looking ahead, they envisage a future in which robotic processes automatically apply ML techniques to data harvested across the entire transaction chain, rather than just select parts of the process as at present.

Introducing new approaches to AML in the middle of the current wholesale revolution in banking isn’t easy. Traditional banks are held back from introducing new, digital-first approaches by a cocktail of legacy technology stacks, dwindling IT budgets and poor data quality. Organisations will only be able to fully leverage operational efficiencies once they look at the big picture and begin to think holistically about the role of AML and compliance within the broader framework of digital transformation across the enterprise.

Working together to accelerate success

In the battle against money laundering, partnerships hold the key. Financial services providers of all types must now consider both national and international collaborations, sharing data, and approaches to combat increasingly sophisticated and international criminal organisations. 

Pooling data is a vital element of successful AI and ML but it does require that the data is clean, well-labelled and from the right sources. That data must then be managed and interpreted in the right way. Almost one in four (24%) of our respondents cited poor-quality data as a key concern for the success of their IT strategy. And they estimated that up to 15% of real-time transactions are being blocked owing to poor data on either recipients or transaction initiators. 

For AML processes to be efficient and effective, data quality must improve radically. Cross-industry collaboration is the best way to bring about that change.

Register here to download the full whitepaper, Better by design? Re-thinking AML for a digital age.

Image: Shutterstock