The upgraded fraud tool adds a unified risk score built on Featurespace technology, Visa’s first combined product since its December 2024 takeover
Visa has launched an enhanced version of A2A Protect, its account-to-account fraud tool, adding a unified fraud score built on technology from Featurespace, the AI firm it bought in December 2024. The score is Visa’s first in-market product to combine the two companies’ systems since the roughly $950m acquisition completed.
A2A Protect covers account-to-account transfers rather than card payments, extending Visa’s fraud tools beyond the card business it is known for. Fraudulent credit transfers cost European consumers and businesses €2.5bn ($2.9bn) in 2024, according to the company.
The tool flags suspect transfers in real time, before money leaves a customer’s account. Global account-to-account transactions are forecast to exceed 5.8trn by 2028, up 160% on 2024, according to the digital payments technology company.
Visa, Featurespace first combined product reaches market
The unified fraud score is the first offering to reach the market from the global card network’s integration of Featurespace, announced in September 2024 and completed that December for about $950m.
Visa placed the Cambridge company in its Risk and Identity Solutions unit and said its models would be built through its existing fraud and risk-scoring range.
Featurespace made its name on adaptive behavioural analytics, machine-learning models that score each transaction against a customer’s normal activity and flag deviations as they happen.
The tool uses AI and transfer learning to give banks immediate access to global risk signals on account-to-account transactions.
Lenders receive intelligence from the first day rather than waiting months for models to learn from their own data, and without joining a consortium of other banks.
Results from an early deployment
One major European bank using A2A Protect reduced over 50% more fraud and cut unnecessary alerts by more than 40%. The company said the tool raised fraud detection by 75% in the first six months after deployment.
Visa said that if similar results were achieved across the industry, around €1.25bn could be detected before funds leave accounts. The figure is a projection, and Visa did not set out the assumptions behind it.
“Fraudsters move fast across payment types, and financial institutions need risk insights just as quickly, without slowing down legitimate payments,” said James Mirfin, Head of Risk and Security at Visa. He said the tool pairs Visa’s network data with Featurespace’s technology to give banks “a powerful new layer of protection” that detects more fraud earlier.
Banks that opt into network-level intelligence sharing receive alerts on emerging scam hotspots and coordinated fraud, patterns Visa said individual institutions find hard to spot alone. The wider group can then respond faster to new threats.
“Fraud is evolving rapidly, and banks need technology that can evolve even faster,” said Mandy Lamb, Head of Value-Added Services at Visa Europe. “We’re giving financial institutions a smart way to identify emerging threats and respond with greater precision, helping them stay ahead in an increasingly complex fraud landscape.”
A2A Protect connects to banks’ existing systems through a single API, which Visa said cuts implementation time. Each alert carries a plain-language explanation of why a transaction was flagged, intended to help fraud teams act without holding up genuine payments.
UK banks have faced mandatory reimbursement of APP fraud victims since 7 October 2024, with payouts capped at £85,000 per claim and the cost split equally between the sending and receiving firm.
The Payment Systems Regulator (PSR) reported that 88% of money stolen through APP fraud, about £112m, was returned to victims between 7 October 2024 and 30 June 2025.
