Payments giant Mastercard has developed its own specialized AI model called Decision Intelligence Pro to assist thousands of banks in detecting and combatting fraudulent transactions. This advanced AI model, as revealed exclusively to CNBC, enables banks to assess suspicious transactions on the Mastercard network in real-time to determine their legitimacy. Ajay Bhalla, President of Mastercard's Cyber and Intelligence Business Unit, disclosed that the company's proprietary solution utilizes a recurrent neural network, a crucial component of generative AI, developed internally by Mastercard's cybersecurity and anti-fraud teams. Bhalla further explained that the model is built using transformer models and draws on extensive transaction data collected from Mastercard's network. While some open-source technology is employed when necessary, the majority of the AI technology is developed in-house. Mastercard's proprietary algorithm is trained on data derived from approximately 125 billion transactions processed annually on its card network. This data helps the AI comprehend relationships between merchants, resulting in the identification of potential fraudulent transactions. The algorithm leverages the history of a cardholder's merchant visits as an input to determine the likelihood of the transaction being genuine. It then generates pathways within Mastercard's network, akin to a heat-sensing radar, to find an answer in the form of a score.

A higher score indicates conformity with the cardholder's usual behavior, while a lower score suggests deviation from the norm. Remarkably, this entire process occurs within 50 milliseconds. According to Bhalla, Mastercard's transaction decisioning technology can enhance financial institutions' fraud detection rates by an average of 20%, with some cases experiencing improvements of up to 300%. Mastercard has invested over $7 billion in cybersecurity and AI technologies over the past five years, including acquisitions like Swedish cybersecurity firm Baffin Bay Networks. Visa, Mastercard's competitor, has also made significant AI investments, including a $100 million venture fund for generative AI startups. Mastercard expects its algorithm to potentially save banks up to 20% by reducing the costs associated with assessing fraudulent transactions. Furthermore, Bhalla believes that the true potential lies in using Mastercard's technology to identify and predict emerging types of fraud that are currently unknown within the payments ecosystem. By leveraging data from a global network of customers, Mastercard can spot fraud patterns on a global scale. Many companies in the payments and digital banking sector have recently acknowledged the transformative impact of AI on their products, including PayPal, which launched new AI-based products and a one-click checkout feature.