Published on February 1, 2024, 7:38 am
Payments giant Mastercard has developed its own proprietary generative artificial intelligence (AI) model called Decision Intelligence Pro to help banks detect fraudulent transactions. This AI model enables banks in Mastercard’s network to assess suspicious transactions in real-time and determine their legitimacy. Ajay Bhalla, President of Cyber and Intelligence at Mastercard, explained that the company’s AI solution is based on a proprietary recurrent neural network, which is a core part of generative AI. The model was built from scratch by Mastercard’s cybersecurity and anti-fraud teams.
Mastercard leverages transformer models to harness the power of generative AI. While the company utilizes open source technology when necessary, the majority of the technology behind Decision Intelligence Pro is created in-house. The algorithm used by Mastercard’s AI is trained on data from approximately 125 billion annual transactions processed through its card network. This vast amount of data helps the AI understand relationships between merchants and predict where fraudulent transactions may occur.
Instead of relying on textual inputs like language models developed by OpenAI or Google, Mastercard’s algorithm uses an individual cardholder’s merchant visit history as input to determine whether a transaction aligns with their typical behavior. The algorithm then generates pathways through Mastercard’s network to find an answer in the form of a score. A higher score indicates typical behavior for the cardholder, while a lower score suggests suspicious activity.
Mastercard claims that its transaction decisioning technology can improve fraud detection rates for financial institutions by an average of 20%. In some cases, it has even led to improvements of up to 300%. Over the past five years, Mastercard has invested more than $7 billion in cybersecurity and AI technologies, including acquisitions such as Swedish cybersecurity firm Baffin Bay Networks.
By leveraging its global ecosystem and analyzing data from customers around the world, Mastercard hopes its algorithm can identify fraudulent patterns and trends to predict future types of fraud that have not yet emerged in the payments ecosystem. This technology has the potential to significantly reduce costs associated with assessing illegitimate transactions for banks.
Other players in the payments and digital banking industry, such as PayPal, also recognize the transformative power of AI. Recent developments include PayPal’s introduction of new AI-based products and a one-click checkout feature.
Mastercard’s Decision Intelligence Pro demonstrates its commitment to utilizing AI and cybersecurity technologies to enhance fraud detection capabilities for banks. As the company continues to refine and develop its proprietary AI model, the industry can expect more innovations that improve transaction security and protect consumers from fraudulent activity.