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

Mastercard has jumped into the generative AI race by introducing a proprietary generative AI model called Decision Intelligence Pro, to improve fraud detection within its network of banks. The algorithm uses “heat-sensing fraud patterns,” assigning scores based on deviations from a cardholder's typical behavior.

Powered by a recurrent neural network developed in-house, the model uses transformer models to leverage the capabilities of generative AI. Mastercard's algorithm, trained on data from approximately 125 billion annual transactions, focuses on understanding relationships between merchants rather than textual inputs.

It predicts potential fraudulent transactions by analyzing a cardholder's historical merchant visits, generating pathways through Mastercard's network to assess the legitimacy of a transaction in just 50 milliseconds.

According to Ajay Bhalla’s (Mastercard’s President of Cyber and Intelligence Business Unit) interview with CNBC, this proprietary technology can enhance financial institutions' fraud detection rates by an average of 20%, with some cases achieving improvements of up to 300%.

“With generative AI we are transforming the speed and accuracy of our anti-fraud solutions, deflecting the efforts of criminals, and protecting banks and their customers. Supercharging our algorithm will improve our ability to anticipate the next potential fraudulent event, instilling trust into every interaction,” he added.

Mastercard has invested over US$ 7 billion in cybersecurity and AI technologies in the last five years, including acquisitions like the purchase of Swedish cybersecurity firm Baffin Bay Networks in March 2023. The company envisions the technology's real strength lies in identifying fraudulent patterns and trends globally, offering insights into emerging types of fraud not currently known within the payments ecosystem.