Mastercard has unveiled a powerful new AI fraud-fighter with serious crime-busting chops. Developed entirely in-house, the company’s new “Decision Intelligence” tech can spot fraudulent transactions with uncanny speed and accuracy.
“We built customized transformer models to harness the power of generative AI,” explained Ajay Bhalla, president of Mastercard’s Cyber and Intelligence unit. Bhalla’s teams leverage Mastercard’s exclusive visibility into over 125 billion annual purchases to train the AI.
Unlike text-focused systems like GPT-3, Mastercard’s algorithm uses actual customer purchase histories as prompts. It quickly generates relationship maps between merchants, with lower scores indicating odd or suspicious activity compared to normal habits.
This rapid transaction scoring allows banks to spot likely fraud and deny charges in under 50 milliseconds – before purchases finalize. And the AI is proving mighty effective at sniffing out criminals.
“In some cases, our new AI has improved fraud detection rates by up to 300%,” said an enthusiastic Bhalla. On average, banks see a 20% boost to savings from fraudulent charges.
Over the past five years, Mastercard has funneled over $7 billion into cybersecurity and next-gen tech. The company sees AI promises in both blocking current fraud as well as identifying emerging threat patterns.
“Because we see global data across our network, it helps us detect cross-ecosystem fraud other systems would miss,” Bhalla added.
With digital banking and eCommerce exploding, Mastercard is betting big on AI to fight an exponentially growing fraud landscape. And with early wins already under its belt, this payments giant seems poised to stay several steps ahead of cybercriminals threatening its nearly 3 billion cardholders.
