Private cycle detection in financial transactions

Abstract

Money laundering is the process of hiding the origin of funds obtained through illicit activities. It is a major problem that has significant impacts on the global financial system and undermines the integrity of financial institutions. To combat this, the Dutch government planning to make it easier for banks to share data to improve the detection of money laundering. However, this approach raises concerns about privacy, as it would allow banks to share sensitive financial information with other banks and institutions. A way to allow banks to still detect money laundering using other banks' data, but without having to share the data would be through the use of multi-party computation. In this work we propose a privacy preserving distributed cycle detection protocol which is meant to find short cycles in financial transactions to help detect money laundering without compromising the privacy of the customers at the bank. Finally, we show that our protocol is significantly faster at detecting short cycles in large financial graphs than current state-of-the-art multi-party computation protocols.