Mastercard debuts a GenAI model, boosting fraud detection by up to 300%.
Mastercard has unveiled a new GenAI model named Decision Intelligence Pro, designed to significantly enhance fraud detection capabilities for banks within its network. This proprietary model employs a recurrent neural network, a pivotal element of GenAI technology, developed in-house by Mastercard’s dedicated cybersecurity and anti-fraud teams. With a focus on real-time transaction assessment, this innovation promises to identify and eliminate fraudulent activities with unprecedented efficiency.
Ajay Bhalla, President of Mastercard’s cyber and intelligence business unit, highlighted the model’s reliance on transformer models to leverage GenAI’s full potential. The model is uniquely trained on data from the 125 billion transactions processed annually through Mastercard’s network, enabling it to predict fraudulent transactions with remarkable accuracy by analyzing merchant relationships rather than textual data.
Unlike conventional methods, Decision Intelligence Pro interprets the history of a cardholder’s transactions with merchants as a prompt to assess transaction legitimacy. It generates a score based on the cardholder’s usual behavior patterns, all within a mere 50 milliseconds. This swift and innovative approach has proven to lead to an average improvement of 20% in fraud detection rates, with some cases witnessing enhancements of up to 300%.
Mastercard’s investment of over $7 billion in cybersecurity and AI over the past five years underscores its commitment to combating fraud through technological innovation. This includes strategic acquisitions aimed at bolstering its cybersecurity capabilities. The company’s forward-thinking strategy not only reduces financial institutions’ operational costs but also pioneers the prediction of emergent fraud types, thanks to its global transaction data insights.