BIS concludes Project Aurora, a proof of concept based on the use of data, technology and collaboration to combat money laundering across institutions and borders

Project AuroraMay 2023The power of data, technology and collaboration to combat money laundering across institutions and borders Project Aurora: The power of data, technology and collaboration to combat money laundering. 3 BISIH Unrestricted Unrestricted Restricted Public Preface: How to read this report For an overview of the project, useful background and key findings read: • Chapter 2: Executive summary • Chapter 3: Introduction • Chapter 4: Project Aurora – proof of concept • Chapter 6: Conclusion For more supporting details read: • Chapter 5: Further considerations • Annex A: Trends and opportunities For other details read: • Annexes B - E Project Aurora was delivered in partnership with: Publication date: May 2023. © Bank for International Settlements 2023. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated. Project Aurora: The power of data, technology and collaboration to combat money laundering. 4 BISIH Unrestricted Unrestricted Restricted Public Contents 1. Acronyms, abbreviations and definitions 7 2. Executive summary 10 1.1 Background 10 1.2 Data, technology and innovation 11 1.3 Project Aurora 12 1.4 Findings and key takeaways 12 1.4.1 A holistic view of payments data unveils money laundering networks 13 1.4.2 Behavioural monitoring and privacy enhancing technologies could be a game changer for AML efforts 13 1.4.3 Leveraging Project Aurora 14 3 Introduction 16 3.1 What is money laundering? 16 3.2 The money laundering process 17 3.3 AML monitoring and analysis today 18 3.3.1 Rule-based monitoring systems 18 3.3.2 Behavioural monitoring systems 19 3.3.3 Defensive reporting and de-risking consequences 19 3.4 Challenges facing AML efforts 20 3.5 Technology 22 3.5.1 Privacy-enhancing technologies (PETs) 22 3.5.2 Graph data structures 23 3.5.3 Machine learning 23 3.5.4 Network analysis 24 3.6 Summary of trends and opportunities 24 4. Project Aurora – proof of concept 26 4.1 Objectives and scope 26 4.1.1 Objectives 26 4.1.2 Scope 27 4.2 Part A: Synthetic data generation 29 4.2.1 Purpose of generating synthetic data 29 4.2.2 Generating the synthetic data 30 4.2.3 Leveraging a three-step approach to generate synthetic data 31 Project Aurora: The power of data, technology and collaboration to combat money laundering. 5 BISIH Unrestricted Unrestricted Restricted Public 4.2.4 Constructing money laundering activities in the synthetic data set 34 4.3 Part B: Application of machine learning to the synthetic data set 39 4.3.1 Machine learning models 39 4.3.2 Testing the models 41 4.3.3 Results 42 4.3.4 Summary 46 4.4 Part C: Testing privacy-enhancing technologies for AML 48 4.4.1 Privacy-enhancing technologies explored 48 4.4.2 Testing a combination of privacy-enhancing technologies in four different collaborative analytics and learning arrangements 49 4.4.3 Results: applying machine learning models in combination with privacy-enhancing technologies 55 4.4.4 Privacy evaluation of PETs when encrypting transaction data. 61 4.4.5 Summary 62 5. Further considerations 65 5.1 Data 65 5.1.1 Additional data and money laundering typologies 65 5.1.2 Real-world data are crucial for understanding the feasibility and impact 66 5.1.3 Limitations of payments data and the need for other data 66 5.1.4 Data protection 67 5.2 Technology 67 5.2.1 Technical challenges with CAL arrangements 67 5.2.2 Machine readable typologies that facilitate information sharing 68 5.2.3 Explainability 71 5.3 Looking ahead 71 5.3.1 Instant payment systems, CBDC systems and financial crime 71 5.3.2 Legal and regulatory considerations 72 6. Conclusion 74 7. Annex A: Trends and opportunities 77 7.1 Standardisation, transparency and harmonisation in payments 77 7.1.1 G20 roadmap for enhancing cross-border payments 77 7.1.2 ISO 20022 harmonisation 77 7.1.3 Data standards for legal entity identification and beneficial ownership 78 7.1.4 The Wolfsberg Group Payment Transparency Standards 79 Project Aurora: The power of data, technology and collaboration to combat money laundering. 6 BISIH Unrestricted Unrestricted Restricted Public 7.2 Transaction monitoring utilities 79 7.2.1 TMU example: Transaction Monitoring Netherlands 81 7.3 Instant payment systems and potential CBDC systems 81 7.4 Public blockchains used for payments 82 8. Annex B: Machine learning in this PoC 83 8.1 Machine learning training, validation and evaluation 83 8.2 Machine learning model feature engineering 84 9. Annex C: Privacy-enhancing technologies 85 9.1 Overview of PETs 85 9.1.1 Homomorphic encryption 85 9.1.2 Local differential privacy 85 9.1.3 Federated learning 86 9.1.4 Other privacy-enhancing technologies 86 9.2 Application of PETs 87 10. Annex D: Questions to support real-world pilots 91 10.1 Objectives, performance monitoring and scope 91 10.2 Data and analysis 91 10.3 Post-pilot questions 93 11. Annex E: Additional acronyms and definitions 94 12. References 96 13. Acknowledgements 100 Project