How AI Is Transforming Fraud Detection

For most, there’s a positive correlation between time spent at home and money spent online. And today, Quarantine x Black Friday have teamed up for the ultimate collab.

But the more your payment details are being sent back and forth in cyberspace, the more opportunities a fraudster has to intercept them. In Q1 2020, credit card fraud reports were up 104% from 2019. (From Q1 2018 to 2019, reports rose just 25%.)

Teams of human analysts can only go so far. As of January 2019, 45% of financial institutions said investigations take too long to complete, and 40% cited too many false positives.

• That’s where AI comes in. Financial institutions (FIs) are banking on the tech to save them major $$ on stolen funds.

The perks

Biometrics: “The hardest thing to spoof is human behavior,” Aarti Borkar, VP of product and strategy at IBM Security, told us. That’s why FIs use AI tools to create biometric profiles that note the times of day you typically shop, the speed at which you typically type your password, etc.

• $217 billion: That’s the amount that AI could save banks by 2030 in areas like authentication, identity verification, compliance, and other data processing functions, according to data from Autonomous Next.

Speed and accuracy: If fraudsters aren’t caught in real-time, investigations can be too difficult and expensive to justify. But slowing down a transaction in order to verify it could cost companies business, making machine learning’s speed critical.

• “The volume of data that can be processed by that AI system in that split second between you hitting ‘enter’ on the password and it figuring out that ‘you’ may not be ‘you’ is where [fraud] is prevented,” says Borkar. “There’s no way to have a human being do that.”

The biggest banks lean heavily on AI tools for fraud prevention, but nearly 95% of all FIs don't. As third-party solutions mature, we’re betting the adoption rate will rise.