Mastercard is set to enhance its real-time fraud detection capabilities with the launch of an innovative AI tool called Decision Intelligence Pro (DI Pro). Developed by the company’s cybersecurity and anti-fraud divisions, DI Pro employs a recurrent neural network to identify fraudulent transactions.
This system is constantly learning, processing billions of Mastercard transactions annually to understand the patterns of cardholders’ purchases, including their relationships with merchants. When a transaction is initiated, the tool evaluates a trillion data points to estimate the likelihood of its authenticity. In less than 50 milliseconds, DI Pro can predict whether a transaction is possibly fraudulent or genuine.
Initial estimates suggest that DI Pro could increase fraud detection rates by a whopping 20% on average, even up to 300% in some scenarios. According to Ajay Bhalla, President of Cyber and Intelligence at Mastercard, generative AI can dramatically improve both the speed and accuracy of fraud detection. He added, “Supercharging our algorithm will improve our ability to anticipate the next potential fraudulent event, instilling trust in every interaction.”
One major benefit of DI Pro will be a substantial reduction in false positives, an issue that can significantly impact operational costs in banks’ fraud prevention departments. DI Pro promises to cut down the number of false positives by over 85%. Since banks often dedicate significant financial and human resources to identifying illegitimate transactions, this reduction could potentially save significant costs by lowering the number of transactions that are falsely flagged as fraudulent. Implementation of DI Pro is expected later this year.