Skip to Content

How a Bank’s AI Credit Card Could Help Solve the Equifax Problem

Now more than ever, consumers are at risk of having their credit cards used fraudulently without their permission, a problem made significantly worse by last year’s massive Equifax hack.

The breach at Equifax, one of the three major consumer credit reporting agencies in the U.S., exposed the sensitive data of more than 145 million people, including credit card numbers, addresses, Social Security numbers, birthdates and more. Such hacks have led to a spike in compromised credit cards: In 2017, more than 14.2 million U.S. credit card numbers were exposed, up 88% from the previous year, according to the Identity Theft Resource Center.

For many Americans, this will mean having to cancel and replace a credit card due to fraud—and then updating their Amazon, Netflix, E-ZPass and myriad other accounts with the new payment information.

Capital One, however, has figured out a way to harness artificial intelligence to keep consumers’ credit card data more secure. The bank will now allow its customers to pay with a separate, virtual credit card number—instead of their real one—every time they make an online purchase, Capital One announced at the South by Southwest conference in Austin, Texas.

Capital One’s artificially intelligent assistant will now generate virtual credit card numbers for consumers.Courtesy of Capital One
Courtesy of Capital One

The technology is a browser extension that runs in the background of a Capital One cardholder’s computer, automatically detecting when the person reaches a checkout page. When that happens, Capital One will generate a virtual credit card number covering that person’s transactions with that specific retailer, and fill in the payment fields on the site. That way, if a retailer were to be hacked, or if the customer identified a fraudulent charge on their bill, Capital One could simply deactivate the compromised credit card alias, instead of replacing the card itself.

“The risks of cybersecurity are growing. We accept that,” says Tom Poole, Capital One’s senior vice president of digital. “At Capital One, we do a great job of solving dollar problems,” he adds, noting that the bank already provides cardholders with protection from fraud. “But the hassle is something we haven’t been able to solve until now.”

Underpinning the virtual credit card capability is the same natural language technology powering Capital One’s artificially intelligent assistant, Eno. Since last year, Capital One customers have been able to merely text the bot their questions, instead of calling customer service. Eno, for example, can process a payment even if a customer just sends a thumbs up emoji, which have accounted for more than half of the payment confirmations sent so far. Indeed, Eno has been asked for a credit card balance more than 2,000 different ways, including by simply texting “$.”

“I don’t know exactly what that means but I like it,” says Ken Dodelin, Capital One’s head of conversational AI. “For decades, we’d been trying to teach customers to speak like bankers and it doesn’t work.”

In addition to handling payments and providing balance and recent transaction info, Eno can tell jokes upon request (“What do you call lending money to a bison? A buff-a-loan”) and also serve up rhymes (“Roses are red, violets are blue, helping you with your money is what I do”). Capital One plans to expand its capabilities to allow Eno to generate virtual credit card numbers via text, for consumers to use even when they are not in front of a computer, and also enable it to address text message inquiries about fraud.

“It’s not lost on us that managing your money is hard,” says Dodelin. “Wouldn’t it be great if there was an intelligent assistant in the background who had your back?”