Visa Unveils First Featurespace-Integrated Fraud Score for Account-to-Account Payments
Visa Unveils First Featurespace-Integrated Fraud Score for Account-to-Account Payments #
Visa has rolled out an upgraded version of its A2A Protect product, bringing its 2024 acquisition of AI specialist Featurespace into a live commercial offering for the first time. The updated platform introduces a unified fraud score to give banks and credit unions faster risk signals when assessing account-to-account transactions.
The announcement, made on Tuesday, comes as A2A payment volumes rise globally. Juniper Research forecasts A2A transaction volumes will grow 83% by 2030. Visa’s press release cited separate projections showing A2A transactions could surpass 5.8 trillion by 2028. Regulators and financial institutions have said the growth is drawing fraudsters to a channel that historically has had fewer protections than card-based payments.
The revamped A2A Protect uses artificial intelligence and transfer learning to generate real-time risk assessments the moment a payment is initiated, without requiring financial institutions to first build large repositories of local transaction data. Visa says this allows the tool to deliver results immediately upon deployment, which it positions as useful for banks entering real-time payment rails for the first time. According to Visa, early results show the tool can increase fraud detection by 75% within the first six months of use, reduce overall fraud losses by more than 50%, and cut unnecessary fraud alerts, which block legitimate payments and frustrate customers, by more than 40%.
For institutions that opt into network-level data sharing, the platform can also surface emerging scam hotspots and coordinated fraud patterns that individual banks would typically be unable to detect on their own.
James Mirfin, Visa’s head of risk and security solutions, said the upgrade responds to fraudsters’ habit of moving quickly across payment types. “Fraudsters move fast across payment types, and financial institutions need risk insights just as quickly, without slowing down legitimate payments,” Mirfin said in the company’s announcement.
Visa originally introduced A2A Protect in 2024, the same year it completed its acquisition of Cambridge-based Featurespace, whose Scam Detect technology underpins the new unified scoring layer. The latest release is the first full commercial integration of that acquisition. The deal had drawn scrutiny at the time over Visa’s expanding footprint in bank-facing fraud infrastructure beyond its traditional card network.
The product is available to banks and credit unions via a single API integration, which Visa says is designed to minimise deployment complexity.
Copy link