Swift has partnered with 13 global banks to demonstrate how artificial intelligence and privacy-enhancing technologies (PETs) could help the industry tackle fraud in cross-border payments. In trials involving ten million artificial transactions, the collaborative model was twice as effective in real-time fraud detection compared to models trained on data from a single institution.
By leveraging PETs, participating institutions were able to securely share artificial transaction data while maintaining end-to-end privacy and security. This approach enabled real-time verification of suspicious accounts and detection of anomalous activity in a test environment. If adopted at scale, this technology could significantly accelerate the identification of cross-border financial crime networks and reduce the billions spent by the industry on fraud-related costs each year.
Collaborating to fight fraud across borders
Financial crime is a global challenge, costing the financial industry an estimated USD 485 billion in 2023 alone, according to Nasdaq Verafin. Tackling this issue requires more than isolated efforts. It demands collaboration across institutions, borders and technologies.
That’s why we’ve brought together a group of international banks and technology partners to test how PETs and AI could be used to securely share fraud insights and detect suspicious activity in real time.
Two use cases, one goal: better fraud detection
In the first use case, participants used PETs to verify intelligence on suspicious accounts across borders without compromising data privacy. This enabled real-time collaboration between institutions, helping to identify complex financial crime networks faster and prevent fraudulent transactions from being executed.
In the second use case, the team combined PETs with federated learning – an AI technique that trains models locally at each institution without sharing customer data. Using artificial data from ten million test transactions, the model was twice as effective at identifying known fraudulent transactions compared to models trained on data from a single institution.
Building on strong foundations
Swift has long worked with the industry to solve common challenges using the latest technology. Earlier this year, Swift launched an AI-enhanced Payments Controls Service to help small and medium-sized institutions flag suspicious transactions more accurately and take action in real time.
And with more than 50 AI pilots currently underway, Swift continues to explore how emerging technologies can strengthen the global financial ecosystem.
What’s next?
Following the success of these initial experiments, we intend to expand participation before launching a second phase of tests using real transaction data. The aim is to demonstrate the impact of these technologies on real-world fraud and bring the industry one step closer to a scalable, collaborative defence against financial crime.
Participants in the experiments included ANZ, BNY and Intesa Sanpaolo, alongside technology partners such as Google Cloud.
Explore the full results
Take a deeper dive into the pilot’s findings and technical innovations in our latest eBook ‘Fighting fraud through data collaboration’.
What the participants are saying
