Artificial Intelligence

How Mastercard Uses AI To Fight Fraud And Make Better Credit Decisions

Mastercard is harnessing artificial intelligence (AI) in a bid to hit fraudsters hard by searching for emerging patterns of criminal activity before they become major problems, two top executives told Karen Webster during Mastercard’s Virtual Cyber & Risk Summit.

“In many cases, AI is fundamental to scaling and keeping up with the pace of the network transactions that are happening,” said Sudhir Jha, a Mastercard senior vice president and head of Brighterion, an AI company that Mastercard acquired in 2017.

He and Nitendra Rajput, Mastercard’s vice president of product development and head of the company’s “AI Garage,” said that in many cases, AI is the only way to scale up sufficiently to meet the challenges the company faces with fraud and other business issues.

“It is one of the key components of our scale strategy in terms of how we want to differentiate and provide value to our customers around the world,” Jha said.

Fighting Fraud in a Post-Pandemic World

The two executives said acquirers need to have better fraud management solutions than ever before, because the pandemic has prompted consumers to use credit cards for more online and app-based transactions.

“We were focusing on acquirers before, and we are doubling down on that and providing both in terms of how to manage risky merchants and also how to manage [a] risky transactions portfolio and to ensure that they’re not getting fined for excessive fraud in their network,” Jha said.

He said Mastercard has long used Brighterion’s technology to provide a fraud score for every transaction that occurs on its network, making the information available to issuers as part of anti-fraud efforts. The company also uses AI for authentication, which Jha said allows it to “decline transactions that need to be declined, but also to approve more transactions and provide a much better experience for our users.”

Additionally, AI is combined with the power and capabilities of NuData, Mastercard’s biometrics and behavioral analysis firm, to spot fraud. Mastercard also utilizes a suite of anti-hacking tools called Safety Net, which provides network-level security by looking for unusual behaviors and potential cyberattacks.

AI Also Helps Manage Credit Risk

Jha and Rajput said the company has also started using AI to focus on areas that have taken on increased importance amid the COVID-19 pandemic.

For instance, Mastercard has been using AI to help its banking partners with credit risk management, aiming to provide the right amount of credit to customers — and the smartest collections efforts — in today’s uncertain economic climate.

“You want to be able to manage credit for the lifecycle of the user,” Jha said. “You want to then be able to predict if there’s a delinquency that needs to be managed. And then once you identify [a delinquency], you want to have a way to do collections that can basically be catered toward that particular user in different ways to provide incentives to recover funds.”

He said that task has taken on greater importance during the pandemic because issuers and loan providers have allowed some delinquent borrowers to take three- or six-month payment “holidays” and then pay back their loans.

On the other hand, Jha said the pandemic has also provided opportunities to provide fresh credit to solid customers who will be able to repay their loans and navigate through the pandemic. He said AI can help spot those clients as well. “They are good customers and therefore will be able to repay later,” Jha noted.

But he also acknowledged the balancing act lenders must manage as they deploy these techniques on their credit risk decisions.

“On one hand, you want to help your customers,” he said. “You want to make sure they can navigate through [this] tough time. You want to give them as much leeway as possible so they are able to pay you back and that they’re given all the opportunities to [do so]. But on the other hand, you want to make sure your business is going to balance the risk and not lose tons of money through the delinquency process.”

Using AI to Optimize Internal Operations

Rajput, the head of Mastercard’s AI Garage, said the company is also using artificial intelligence internally to improve its work processes. He said the AI Garage is always looking for areas where AI can help, such as in recruiting.

According to Rajput, Mastercard has begun using AI to identify which job applicants have the specific skills required for a particular role. He added that an AI-specific technique called natural language processing can understand that when someone has experience with “Python,” they’re referring to a programming language and not a snake.

“It can identify which skills are required for which job role and which people are the most [qualified] with these skills,” Rajput said. “Not only that, if this employee wants to go from level one to level X at different job roles, what are the skills that are required for this employee to get there? And what are the courses that would take her there?”

Additionally, Rajput said Mastercard uses AI to predict revenue — not just for the company as a whole, but for different products within the company.

“We’ve been working on identifying ways to predict revenue for Mastercard – and it’s not just Mastercard, but also different geographies within Mastercard, different products within Mastercard,” Rajput noted. “Looking at the past data, can you figure out how the future is going to be from a revenue perspective? This obviously has a significant impact on how a company would take its next steps in terms of making investments or taking the big bets.”

Using AI for Good

Mastercard has also been using its AI skills to help the pandemic situation. For example, the AI Garage team used its expertise in predicting revenue and adapted that to predict the number of COVID cases as part of the White House-supported Kaggle competition. It also used its ability to understand text documents and applied that to yet another Kaggle competition, which was focused on making sense of the over 100,000 research articles published on the coronavirus.

TRENDING RIGHT NOW