Finance
Finance
Secure Multi-Party Computation for Fraud Detection
The Challenge
United Nations estimates indicate that between 2 and 5% of global GDP is laundered every year. We refer by Money Laundering to the act of obscuring financial flow to re-integrate money, that was illegitimately sourced, into the financial system while hiding its origins. This is not only a crime itself but is also often linked to more severe crimes like drug trafficking. Financial institutions are obligated to detect Money Laundering and terror financing patterns and provide reporting to regulatory bodies. However, Money Laundering behavior often happens across different banks. This would make it very valuable for Anti-Money Laundering efforts to run a joint analysis between multiple banks to spot irregular patterns, but privacy laws prevent banks fr om sharing raw customer data.
The Q-FENCE Solution
Under the coordination of the University of Luxemburg, Q-FENCE will create a pilot implementation for a cross-bank Anti-Money Laundering system, relying on techniques from Secure Multi-Party Computation (MPC). MPC allows participating financial institutes to jointly compute on encrypted data from multiple institutions to find suspicious patterns without ever revealing more information about the underlying private details, than is revealed by the jointly computed functionality. For example, such a functionality could return to each bank information about a “risk class” of their own accounts, but no private information about any other banks’ accounts. One technique we envision to use in building this special-use MPC system is (Fully) Homomorphic Encryption, which allows to evaluate low-complexity functionalities on encrypted data.
Location
Luxembourg
Key Features
The pilot will feature a tamper-proof logbook secured by post-quantum signatures to create an audit trail for regulators.
Its MPC functionality will focus on detecting fraud patterns that can only be seen from the data of multiple banks together while preserving confidentiality.
For example, such a pattern could be cycles of sending money through various shell companies to reintroduce it as “legitimate cash flow” to the original account. Our system is therefore to be seen as an addition, not a replacement to current systems ensuring compliance with Anti-Money Laundering (AML) directives.
The Impact
This use case demonstrates how privacy-preserving technologies can enhance cross-bank collaboration in Anti-Money Laundering investigations. By enabling additional secure analytical capabilities on encrypted data, the approach strengthens regulatory compliance while preserving data confidentiality and institutional autonomy.