The Bank for International Settlements (BIS) Innovation Hub brings central banks together from all over the world to experiment and develop financial technology for central banks, and also to share information and provide education. One key topic of discussion for the last three years has been AI and how this technology changes the landscape from a fincrime perspective.
We caught up with Beju Shah, Head of the Nordic Center at BIS Innovation Hub, for his insights.
NFM: How is technology changing the financial crime landscape?
Shah: Cybercriminals are using cutting-edge technologies faster than regulators or the private sector can keep up with. It’s easier than ever for cybercriminals to adopt technology-enabled services like Money Laundering as a Service and Cybercrime as a Service. Technology and AI are reducing the barrier to entry, allowing bad actors to test multiple options and variations of attack vectors rapidly, and criminals are using AI at scale to automate attacks with increased sophistication.
Join us at Stockholm Fintech Week on Feb 11-12!
What are the main challenges in combating these new types of attacks?
Shah: Ultimately, money laundering is an international problem. It’s a borderless payments problem. Tackling the problem successfully requires information sharing and collaborative approaches to analysis across institutions and ultimately across borders.
It’s not currently possible for banks to rapidly track and adapt to evolving and complex typologies—the patterns used by money launderers—across borders because of variances in data protection regulations which, while necessary, prevent banks from sharing data easily.
In a survey we did of about 100 major financial institutions and FinTechs, 72% agreed that a holistic network view of transactions and flows is essential to effectively combat money laundering and fincrime, and 82% agreed that information sharing and collaborative approaches to analysis are ultimately needed.
Join us at Stockholm Fintech Week on Feb 11-12!
What have we learned about effective detection?
Shah: At the BIS Nordic Centre, we ran a highly successful proof of concept called Project Aurora, which used AI, ML, privacy-enhancing technologies, and advanced analytics that demonstrated that, by applying these to analysis, the broadest possible view of data could achieve a three-fold increase in detection of money laundering but also a reduction in false positives of up to 80%. We’ve now moved into the second phase of the project.
Are banks using AI and ML to try and combat sophisticated attacks?
Shah: Although unsupervised machine learning, particularly graph neural networks, seems to be the most effective of all the technologies tested so far, a lot of questions surround it, particularly regarding explainability. About 25% of banks are using or trying it and have found it promising. Over 50% are using or trying network analysis, graph analytics, and anomaly detection.
What progress is being made on cross-border data sharing?
Shah: Europe’s incoming Anti-Money Laundering Regulations have provisions for better data sharing in Article 75. The regulation doesn’t solve everything but it’s a stride in the right direction. Technologies like federated learning could help but there’s a lot more non-technology work that needs to be done to move progress forward. At the very least, starting nationally is a major step forward.
How is BIS Innovation Hub helping?
Shah: BIS Innovation Hub’s purpose is to educate, show the art of the possible with the latest technologies, and move conversations forward. Several projects across our global centres are looking at various aspects of this problem.
We must be mindful that no silver bullet exists. The threat actors are sophisticated and some have immense resources. Leveraging tech and data in new ways to improve the visibility of suspicious networks and increase the criminals’ risk could make it harder for them to profit from such activities, especially by slowing them down sufficiently.