Bank-to-bank messaging network Swift is hailing the results of tests that used a combination of AI and cross-border data sharing to double real time fraud detection in trials involving 13 top tier banks and ten million test transactionson.
Editorial
This content has been selected, created and edited by the Finextra editorial team based upon its relevance and interest to our community.
The experiments used privacy-enhancing technologies (PETs) to enable institutions to securely share fraud insights across borders. In one use case, the PETs enabled participants to verify intelligence on suspicious accounts in real-time, a development which could speed up the time taken to identify complex international financial crime networks and avoid fraudulent transactions being executed.
In a separate use case, participants used a combination of PETs and federated learning - an AI model that ‘visits’ each institution to train on its data locally, so that institutions can collaborate without sharing customer information - to identify anomalous transactions.
Trained on synthetic data from ten million artificial transactions between participants, the model was twice as effective in identifying instances of known frauds as a model trained on a single institution’s dataset.
Fiancial institutions involved in the trials include ANZ, BNY and Intesa Sanpaolo, as well as technology partners including Google Cloud.
Rachel Levi, head of AI at Swift, says: “The industry loses billions to fraud each year, but by enabling the secure sharing of intelligence across borders we’re paving the way for this figure to be significantly reduced, and allowing fraud to be stopped in a matter of minutes, not hours or days.”
Following these successful experiments, Swift intends to expand participation before launching a second phase of tests, which will use real transaction data and seek to demonstrate the technologies’ impact on real-world fraud.
The cooperative has been actively exploring the role AI can play in solving challenges in cross-border payments, and currently has more than 50 use cases across proof of concept, pilots and live usage.
Earlier this year Swift launched an AI-enhanced Payments Controls Service, which helps small and medium-sized financial institutions more accurately flag suspicious transactions so that action can be taken in real time.
