Mastercard Is Using AI to Fight Fraud

Credit card fraud threatens both personal spending as well as your business’ finances. Bad actors use a series of tricks to steal credit card information, and use that data to spend as much as they can before the target catches on and cancels the card.

Like all credit card companies, Mastercard wants to fight back. However, they’re now taking a modern approach, relying on AI models to help banks detect fraudulent transactions.

Mastercard’s “Decision Intelligence Pro”

Mastercard announced its new AI model, Decision Intelligence Pro, to CNBC. The model is designed to help financial institutions analyze suspicious activity as it happens, and determine on the fly whether or not those transactions are false or malicious.

Decision Intelligence Pro is a recurrent neural network, which is an AI model built with memory to follow sequential data. Mastercard also built most of this neural network themselves, using in-house talent from the cybersecurity and anti-fraud teams. The company says it used open-source tech at times, but tried to keep most of it proprietary. That should serve them well when dealing with clever fraudsters.

Mastercard has the data to make this work

An AI is only as good as the data it’s trained on, and Mastercard has that covered. The company trained its recurrent neural network on the 125 billion annual transactions that come through Mastercard. The AI can study the patterns between legitimate transactions between buyers and sellers. After billions of these interactions, it should be able to spot a fraudulent transaction among the legitimate.

In fact, it even bases its analysis on your transaction history, deciding whether the transaction “you” just engaged in was with a seller you would likely buy something from. It leans on its data to assign the transaction a score: The higher the score, the more likely the transaction would happen, and would be legitimate. The lower the score, the less likely the transaction would happen with this cardholder, and so, the higher the chance this transaction is fraudulent. All that happens in about 50 milliseconds. Mastercard prides its model on raising the rate at which fraud is detected by 20%–300%.

Will it work?

Mastercard isn’t the only credit card company looking to use AI to improve fraud detection. Visa has invested $100 million into generative AI startups, which means there’s going to be competition in this space.

But that’s a good thing for all of us, of course: The better credit card companies can make the tech they use to stop fraud, the more generally secure our transactions will become.