Visa launched an enhanced version of A2A Protect, its fraud-detection tool for account-to-account payments, built on technology from Featurespace, the AI fraud-prevention firm Visa acquired in 2024. The new version adds a single unified fraud score meant to flag bank-transfer fraud before money leaves the sender’s account, rather than after the transfer has already cleared. Visa said the tool has increased fraud detection by 75 percent and cut unnecessary fraud alerts by more than 40 percent during deployments over its first six months.
Why It Matters
Account-to-account payments are growing precisely because they skip the card network, which is also what makes them harder to police. Visa’s own projection puts A2A transaction volume at 5.8 trillion dollars by 2028, a 160 percent increase from 2024. James Mirfin, Visa’s head of risk and security solutions, framed the stakes plainly: “Fraudsters move fast across payment types, and financial institutions need risk insights just as quickly, without slowing down legitimate payments.”
The Original Insight
The more interesting move here is architectural, not statistical. By building A2A Protect to work without local model training or a bank-by-bank consortium, Visa is repeating the same strategy it used when it paid 2.4 billion dollars for fraud-detection firm BioCatch: buy the specialist technology, then sell it back to banks as a shared utility rather than compete with each bank’s own fraud stack. Visa has used the same playbook of taking a niche capability and turning it into network-wide infrastructure before, and each repetition makes it harder for any single bank to build a competitive fraud model in-house rather than simply buying Visa’s.
Source: Visa Investor Relations
