Swift has conducted experiments involving 13 global financial institutions, demonstrating that artificial intelligence (AI) combined with privacy-enhancing technologies (PETs) can significantly reduce fraud in international payments by enabling secure cross-border data collaboration. These trials assessed the potential of AI and secure data sharing to combat financial crime.
Swift’s experiments utilized PETs to facilitate the secure sharing of fraud insights among participating institutions. In one use case, PETs enabled real-time verification of intelligence on suspicious accounts, a development that could accelerate the identification of complex international financial crime networks and prevent fraudulent transactions from being executed.
In a separate experiment, institutions employed a combination of PETs and federated learning—an AI model that trains on each institution’s local data without direct sharing of customer information—to detect anomalous transactions. This model, trained on synthetic data from ten million artificial transactions between participants, proved twice as effective in identifying known frauds compared to a model trained on a single institution’s dataset. Rachel Levi, Head of AI at Swift, highlighted the importance of this collaborative approach, stating, “A united, industry-wide fraud defence will always be stronger than one put up by a single institution acting alone.” She added that enabling secure cross-border intelligence sharing could significantly reduce the billions lost to fraud annually, allowing fraud to be stopped in minutes rather than hours or days.
Following these initial successes, Swift plans to broaden participation for a second phase of tests. This next stage will incorporate real transaction data to demonstrate the technologies’ impact on actual fraud cases. Swift has a history of industry collaboration to address common challenges and enhance the speed, efficiency, and security of cross-border transactions, actively exploring over 50 AI use cases across proof-of-concept, pilot, and live implementations.
The financial industry incurred an estimated USD 485 billion in losses due to financial crime in 2023. Earlier this year, Swift introduced an AI-enhanced Payments Controls Service, designed to help small and medium-sized financial institutions more accurately flag suspicious transactions for real-time intervention.
Participating financial institutions in the experiments included ANZ, BNY, and Intesa Sanpaolo, alongside technology partner Google Cloud. David Buckthought, Head of Technology – Payment Services and Digital Assets at ANZ, remarked on the global impact of fraud and the value of industry-wide responses, stating that federated learning enhances detection capabilities and provides banks with a stronger defense. Isabel Schmidt, Executive Platform Owner at BNY, emphasized the paramount importance of security in cross-border payments, noting that the results demonstrate how these tools can uplift the entire ecosystem. Enrico Canna, Head of Anti-Fraud & Customer Protection Centre at Intesa Sanpaolo, commented on how fraud increases friction and costs, underscoring the positive impact of a synergistic approach supported by advanced technologies to enhance ecosystem security.
Swift operates as a global member-owned cooperative, providing secure financial messaging services to over 11,500 banking and securities organizations, market infrastructures, and corporate customers across more than 200 countries and territories. While it does not hold funds or manage accounts, Swift facilitates secure, standardized financial message exchange, supporting global financial flows and trade.
