City partners with Kindred and BetBuddy to explore use of artificial intelligence in anti-money laundering initiatives
A collaboration between City, University of London’s Research Centre for Machine Learning, the online gambling operator Kindred, and BetBuddy, Playtech’s responsible gambling data analytics company, has produced the Interview Stakeholder White Paper, titled, “Raising Standards in Compliance: Application of artificial intelligence to online data to identify anomalous behaviours”.
It represents the first findings from a three-year project which will explore how artificial intelligence can be used in anti-money laundering initiatives. This initial research focused on analysing areas requiring improvement for artificial intelligence to be deployed. The next phase of research will use real-world online gambling data to detect signs of money laundering.
The White Paper is authored by City PhD candidate Charitos Charitou; Simo Dragicevic, CEO of BetBuddy, which is part of Playtech Plc, and City PhD supervisor; and City Professor Artur Garcez, Director of the Research Centre for Machine Learning.
The paper is one of the fruits of the partnership entered into by City with BetBuddy and Kindred in 2017 to sponsor a PhD student over three years to explore the use of deep learning and artificial intelligence (AI) techniques to strengthen anti-money laundering (AML) decision processes across the UK online gambling industry.
Key technical recommendations
Alongside a summary of the main discussions taken from the expert and stakeholder interviews - which included experts from national crime agencies, regulators, trade associations, suppliers, and operators - the paper identifies a number of challenges to the online gambling industry and makes key technical recommendations, some of which should be tackled at industry level, and others which will form the basis of later phases of technical research:
- The development of a single format or technical protocol for submitting Suspicious Transaction Reports (STRs), Suspicious Activity Reports (SARs), and Defence Against Money Laundering (DAML) across jurisdictions that enable operators to submit cases using a consistent system while also providing feedback on submission quality;
- The continuation of efforts to develop a single central database for customers flagged for suspicious gambling activity, to enable enhanced monitoring of flagged customers across industry;
- For threshold checks, exploring the addition of more threshold levels above the current regulatory requirements and the introduction of variable elements in the process;
- Developing more sophisticated and cost-efficient methods to improve ongoing monitoring. This entails building techniques that analyse player’s behaviour below the minimum threshold levels required by regulators, whilst not relying on increased staff numbers to broaden monitoring scope;
- Using data to develop more sophisticated behavioural checks and customer affordability segments to support enhanced source of wealth (SOW) and source of funds (SOF) checks throughout the customer lifecycle for higher spenders, and not just at specific points such as regulatory threshold breaches (e.g. a customer depositing over €2K within a 24-hour period);
- Investing in the modernisation and simplification of know your customer (KYC) and SOF processes, and using it as an opportunity to build a closer customer relationship and to build trust in the gambling brand, - and not view it as an administrative or ‘check box’ compliance process;
Professor Artur Garcez, Head of City’s Research Centre for Machine Learning, said:
We are looking forward to starting the next phase of the research using real-world online gambling data. Some of the techniques available today to detect anomalous behaviour, including state-of-the-art recurrent neural networks, have shown promising results in the analysis of such complex streaming data.
Simo Dragicevic, CEO of BetBuddy, added:
“This initial phase has been important in obtaining stakeholder validation of where areas for improvement exist in AML monitoring. It is clear that whilst this is a very complex challenge, expectations for continuous improvement and investment in research and development using new technologies is high amongst stakeholders.”
Maris Bonello, Head of Player Sustainability and Integrity Analytics at Kindred Group, added:
“Collaboration across research, regulators, operators and other partners is crucial if we are to improve techniques and tools to fight fraudulent behaviour across digital platforms. We believe that engaging in research such as this will encourage more transparency and help bridge the work done by operators and academia.”
Participating companies and organisations involved in the Interview Stakeholder White Paper included: Kindred Group, Remote Gambling Association, the Malta Gaming Authority, Financial Intelligence Unit, Malta, Playtech Plc, EPIC Risk Management, and the UK Gambling Commission.