Visa has announced an upgraded version of A2A Protect, its fraud-detection service for account-to-account payments. The Kenya release is dated September 8, 2026. The update adds a single fraud score built with technology from Featurespace, the Cambridge-founded fraud analytics company Visa acquired in December 2024.
This is Visa’s first in-market product integration of Featurespace technology, according to the company. The aim is to help banks decide whether a transfer looks suspicious before the money leaves a customer’s account.
A2A Protect itself predates the Featurespace acquisition. Visa announced Visa Protect for A2A Payments in 2024, when it said the service was live in Latin America and being piloted in the UK. The September 2026 update is therefore an upgrade to an existing fraud product, with Featurespace’s technology now folded into the scoring system.
What A2A Protect actually does
Account-to-account, or A2A, payments move money directly between payment accounts instead of using a card transaction. That includes many bank-to-bank transfers and real-time payment systems.
A2A Protect sits alongside a bank’s existing fraud systems. A financial institution sends transaction data through a single application programming interface, or API. Visa returns a risk score and a plain-language explanation of why the payment may be suspicious.
The system also uses Featurespace’s behavioural analytics. Featurespace builds models of what normal activity looks like for individual customers and then looks for transactions that do not fit those patterns. Visa says its newer approach also uses transfer learning so a bank can start with fraud intelligence from existing models instead of waiting months to build a useful model only from its own transaction history.
Banks can separately opt into network-level intelligence sharing. That gives the scoring system more signals from across participating institutions and can help identify coordinated scams or fraud patterns that may be difficult to see from one bank’s data alone.
Visa says it can catch more fraud with fewer false alarms
Visa’s strongest public evidence comes from work with Pay.UK. On its A2A Protect product page, Visa says the system identified 54% of fraud and authorised push payment scam value beyond what participating banks’ own systems had already detected. Visa also reports a 40% reduction in false positives when Visa data was added.
False positives are genuine transactions incorrectly flagged as suspicious. Too many of them can block customers and force fraud teams to spend time reviewing harmless transfers.
The figures are Visa’s reported results from the Pay.UK work. They should not be read as a guarantee that every bank will see the same performance.
In June, we covered Visa’s Stay Secure 2026 survey, which found that 37% of Kenyan respondents said they had experienced a financial scam in the previous 12 months. In the same survey, 67% said real-time alerts from a bank or payment app would make them feel safer.
What this means for Kenya
Visa’s September 8 Kenya release does not name a Kenyan bank that is using the upgraded A2A Protect, nor does it give a local deployment timetable. So there is no customer-facing change to point to in a Kenyan banking app from this announcement alone.
Visa is selling fraud detection for transfers that do not need to run on Visa’s card network. Featurespace gives it behavioural models for those payments, while Visa adds broader payment intelligence and a single risk score.
For Kenyan customers, the practical effect will depend on whether local banks and payment providers adopt the service. Until Visa names those institutions, this is a new fraud tool available to financial institutions rather than a new feature consumers can use directly.
Join the discussion