Vyntra: A Behaviour-Led Blueprint to Stop Payment Scams

At Sibos 2025, Alexandre Guidou, Associate Sales Director for Europe at Vyntra, explains how banks can get ahead of external fraud. Guidou highlights the hardest patterns to stop: payment, investment and romance scams initiated by genuine customers; technical attacks like account takeover; and smurfing, where criminals split transfers to dodge thresholds, issues that cut across fraud and AML.

Vyntra’s fix is a proactive, AI-led approach that learns each customer’s normal behaviour and flags anomalies in real time. Instead of static rules, banks should risk-score every payment and trigger fresh due diligence when something looks off. He outlines a two-pronged data strategy: first, use first-party data to profile habits and compare customers to peer groups; second, enrich with third-party intelligence, shared lists of suspicious beneficiaries and scheme-level risk signals, then orchestrate both to lift detection performance.

The bigger hurdle, however, is organisational. Guidou urges breaking silos between fraud and compliance so analytics and investigations share signals and context. Let models learn from each other’s flags and give analysts a single view, especially for money-mule detection, which spans both domains. Connecting fraud and AML, move to behaviour-led, real-time controls, and fuse internal and external data stop losses before funds leave the account.