Swift AI launches cross-border fraud detection with AI

By Vriti Gothi

Swift AI has collaborated with 13 major international banks and demonstrated a breakthrough in real-time fraud detection through pioneering experiments that leverage AI to secure cross-border data collaboration. The trial, which includes 10 million synthetic transactions, shows that a coordinated, technology-driven approach can significantly enhance global financial crime prevention.

In this experiment, we deployed advanced privacy-enhancing technology (PET) to allow the institution to share transaction intelligence while maintaining strict end-to-end privacy and security. In one use case, the pet allowed participants to verify suspicious accounts in real time, accelerate the identification of complex international fraud networks, and prevent fraudulent transactions before they could be implemented.

In the second, more advanced scenario, we combined Federated Learning, an AI methodology that locally trains models at each institution without revealing customer data, with pets to detect unusual transactions. When trained on synthetic datasets, this joint AI model doubled the effectiveness of detection compared to models trained alone in individual banks.

Rachel Levy, head of Swift's AI; “These experiments demonstrate Swift's convening power as a trustworthy cooperative at the heart of global finance. The unified industry-wide fraud defense is stronger than that which has always been held back by a single institution acting alone.

Following these successful trials, Swift plans to expand participation and begin the second phase of the experiment using real transaction data, with the aim of verifying the practical impact of these technologies in live environments.

Swift has long been at the forefront of innovation in the financial industry, combining advanced technology and collaborative models to increase the speed, efficiency and security of cross-border payments. Currently, Swift explores AI applications across more than 50 use cases, spanning concepts, pilot projects, and proof of live deployments. Earlier this year, the organization introduced the Ai-Enhanced Payments Controls Service, allowing small and medium-sized financial institutions to flag suspicious transactions in real time.

Participants in the exam included ANZ, BNY, and Intesa Sanpaolo, along with technology partners such as Google Cloud.