SARS, nets, and autoencoders:A road trip through AI and financial crime
Abstract
This inaugural lecture examines how artificial intelligence can strengthen the response to financial economic crime and identifies three interlinked priorities for sustainable progress. It presents evidence from research on behavioural fingerprints, autoencoders, graph neural networks and foundation models, and discusses operational uses of large language models, synthetic datasets and federated learning. The lecture outlines a practical vision for an automated feedback loop that turns investigative intelligence into rapid, testable detection hypotheses. The main points are: - Collaboration is essential: academia, industry and public partners must share data, tools and aligned priorities to build realistic, deployable solutions. - AI can offer real benefits: improved detection, smarter investigations and continuous learning loops can reduce false positives and help organisations adapt to evolving criminal tactics. - Do not hesitate to innovate: use available data, run experiments (including synthetic and federated approaches), accept small failures, and reward learning rather than only immediate success. Together, these steps can make anti-money-laundering efforts more effective, efficient and adaptive.
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.