Mastercard's DI Pro: How AI Fraud Models Achieve Real-Time Transaction Security in Under 300 Milliseconds
Mastercard's Decision Intelligence Pro (DI Pro) platform leverages sophisticated AI models, specifically a recurrent neural network (RNN) with an "inverse recommender" architecture, to detect fraudulent transactions in real time. Processing approximately 160 billion transactions annually and up to 70,000 transactions per second during peak times, DI Pro assesses the risk of individual transactions within milliseconds. This speed is crucial as transactions flow from consumer to network to issuing bank, typically completing in under 300 milliseconds. Unlike traditional anomaly detection, DI Pro identifies suspicious transactions that mimic legitimate consumer behavior by performing a pattern completion exercise to understand merchant relationships. Johan Gerber, Mastercard’s EVP of security solutions, emphasizes the platform's ability to provide precise, contextualized risk scores to issuing banks, enabling better approve-or-decline decisions.
Fraud protection is a constant battle against immense scale. Mastercard's network, for example, handles roughly 160 billion transactions each year, experiencing surges of up to 70,000 transactions per second during peak periods, such as the December holidays. Identifying fraudulent purchases amidst this volume, without generating excessive false alarms, is an incredibly challenging task, which historically has allowed fraudsters to exploit the system. However, advanced AI models are now capable of scrutinizing individual transactions, pinpointing suspicious activity within milliseconds. This capability is central to Mastercard’s flagship fraud platform, Decision Intelligence Pro (DI Pro).
Johan Gerber, Mastercard’s EVP of security solutions, explained in a recent VB Beyond the Pilot podcast that "DI Pro is specifically looking at each transaction and the risk associated with it." He added, "The fundamental problem we're trying to solve here is assessing in real time." DI Pro was engineered with a focus on low latency and high speed. From the moment a consumer initiates a transaction, whether by tapping a card or clicking "buy," the transaction moves through Mastercard’s orchestration layer, onto the network, and then to the issuing bank. This entire process typically concludes in less than 300 milliseconds.
While the ultimate decision to approve or decline a transaction rests with the bank, the quality of that decision largely depends on Mastercard’s capacity to deliver an accurate, contextualized risk score indicating the likelihood of fraud. A key challenge in this process is that the system isn't merely searching for anomalies; instead, it's designed to identify transactions that intentionally resemble legitimate consumer behavior. At its core, DI Pro utilizes a recurrent neural network (RNN) that Mastercard refers to as an "inverse recommender" architecture. This innovative approach frames fraud detection as a recommendation problem. The RNN executes a pattern completion exercise to discern how different merchants relate to one another. As Gerber elaborated, it considers factors like: "Here's where they've been before, here's where they are right now."