Regulators need to establish a “minimum set of data” for banks to share, which must effectively support fraud detection and prevention, according to Michele Gentile, head of group correspondent banking for APAC at UniCredit, during a session at Sibos.
The fight against fraud is increasingly vital as financial criminals utilize advanced technologies, including generative AI (GenAI), and exploit real-time payment systems, costing the global economy an estimated $485 billion in 2023.
At Sibos on its second day, a panel discussion focused on the importance of collaborative data sharing in addressing fraud. Participants included Sergio Antonio Dalla Riva from Intesa San Paolo, Ute Kohl from Deutsche Bank, and Gentile, who examined Swift’s Federated Learning AI initiative. Launched at the previous year’s event, this initiative unites banks and tech providers to explore how AI-driven data collaboration can significantly mitigate fraud while tackling issues of data governance and privacy.
In a discussion led by moderator Amit Mahajan, head of innovation at SWIFT, the panel reviewed achievements over the past year and discussed the next steps toward enhancing the global banking network’s ability to combat financial crime.
Gentile illustrated the necessity for regulators to define a baseline for fraud data sharing using the example of money mules. He pointed out that verifying payees can be strengthened if issuers and acquirers can exchange account information. However, current privacy laws often restrict such communication. He emphasized the need for mandatory regulations to ensure essential data exchange among banks.
Mahajan noted that Swift is exploring the use of “synthetic data” to test these concepts before implementing real data, stressing the importance of both the “what” and “how” of sharing fraud information.
Despite the potential benefits of a systematic fraud data exchange, the panel acknowledged the financial sector’s reluctance stemming from privacy concerns. Riva emphasized the goal of protecting customers, while Gentile recognized the challenge of ensuring data security. Both advocated for a balance between data sharing and customer protection, with Gentile calling for mandatory sharing of relevant information among payment service providers to counteract fraud.
Kohl suggested that concerns could be alleviated as fraud patterns and insights can be shared without disclosing sensitive data. The discussion broadened to include the necessity for collaboration across sectors, particularly with local authorities, highlighting that different entities manage and analyze data in contrasting ways.
The panel also explored the role of artificial intelligence in combating financial fraud. While AI is often considered a potential solution, Mahajan questioned how prudent it is to rely on a relatively immature technology. Kohl warned that AI’s effectiveness depends on human oversight, acknowledging the evolving nature of machine learning and algorithm development.
Gentile highlighted the promise of federated machine learning, which allows AI training without direct access to data, contingent on consent within the banking sector. He stressed the urgency to act decisively against the pervasive threat of scams, noting that without swift action, the industry risks falling behind.