Singapore’s UOB Integrates Machine Learning in AML Systems

Collaboration with regtech firm Tookitaki has enabled deeper and broader analyses of data sets which can be applied to all processes within AML framework.

Singapore’s UOB (United Overseas Bank) has announced the use of machine learning in its fight against money laundering, following a partnership with Tookitaki to develop the technology.

Using Tookitaki’s Anti-Money Laundering Suite (AMLS), UOB co-created a number of machine learning features which enable deeper and broader analyses of any set of data for greater accuracy beyond what is provided by existing rules-based AML (anti-money laundering) systems.

The solution will allow the bank to draw out more precise information, more quickly, to prevent and to detect suspicious money laundering activities, and can be applied to all processes within the AML framework. This represents an improvement over the industry norm, which is to use multiple systems to analyse subsets of the same set of data for each process.

“With this integrated solution supplementing the Bank’s existing AML systems, UOB is able to make sharper, smarter and swifter detection of high-risk individuals and companies and suspicious activities,” said UOB. “This is essential given the volume, value and velocity of transactions that flow through the Bank’s systems.”

UOB has so far applied the solution for two main processes within the AML framework. In name screening, it identifies high-risk individuals and entities based on internal and external watch lists, enabling the bank to more accurately assess the risk involved in banking them, and to guard against AML activities. In transaction monitoring, the solution flags suspicious transactions for investigation.

The AMLS additionally creates a smart rule when it spots a pattern of suspicious activity, enabling machine learning models to detect similar patterns for future alerts. Over time, this leads to more accurate tracking, saving employees time that is better spent conducting investigations or focusing on other cases.

“The use of RegTech such as Tookitaki’s AMLS enables us to augment our ability to identify actionable alerts and to minimise false positives. These sharpen the accuracy and effectiveness of our AML risk management,” said UOB Head of Group Compliance Victor Ngo.

“The six-month pilot has shown how AMLS can enhance our processes and over the next six months, we will continue to optimise AMLS’ machine learning algorithms by adding new transactional data into the database. We will then implement the solution across the entire AML framework over time.”

During the six-month pilot, UOB experience a 60 percent reduction in false positives for individual names and a and 50 percent reduction for corporate names. Transaction monitoring saw a five percent increase in true positives and a 40 percent drop in false positives.

To Top