Mastercard Launches Decision Intelligence Pro, a New Generative AI Model for Fraud Detection

Mastercard has launched Decision Intelligence Pro, a proprietary generative AI model designed to dramatically improve fraud detection by scanning one trillion data points to assess transaction authenticity in real time.

The new system analyses relationships between entities involved in a transaction - including merchants, account holders, and devices - to generate a risk score that predicts whether a transaction is genuine or fraudulent. Mastercard said the model can improve fraud detection rates by up to 300% compared to existing methods, while simultaneously reducing false positives that inconvenience legitimate cardholders.

Decision Intelligence Pro builds on Mastercard's existing Decision Intelligence platform, which has been used by issuing banks worldwide to make real-time authorisation decisions. The generative AI upgrade adds the ability to identify complex, previously undetectable fraud patterns by examining how transactions relate to broader networks of activity rather than evaluating each transaction in isolation.

The model was trained on anonymised transaction data from Mastercard's global network, which processes billions of transactions annually. The company said no personal data is used in the training process, and the system operates within existing data privacy frameworks.

The launch comes as card fraud continues to evolve in sophistication, with criminals increasingly using AI tools to generate synthetic identities and conduct social engineering attacks. Mastercard's response positions generative AI as a defensive technology, turning the same capabilities being exploited by fraudsters into tools for detection and prevention.