How BioCatch Uses Behavioral Biometrics To Catch Fraud With CEO Gadi Mazor | Forbes

8200 is one of the largest units in the army. The equivalent is probably just like the US NSA, but unlike the US NSA, it's not a professional service. It's basically kids at the end of the age of 18 that get recruited, get tested, go through all kinds of screening when they are 17 or 18. They get to those units. They usually serve not the mandatory three years but five years in very specific units within that large unit.

Without even knowing that they can succeed. And that's the type of training that I think then led many of the alumni to start their own companies and basically not be afraid of big challenges after the unit.

So I left the service when I was just finishing my master's. My master's were in mathematics, and I was considering—I was actually going to Stanford, MIT, and Harvard to look at where I wanted to do a PhD. Then a friend of mine and I thought about an issue that would be nice to solve, and we started the company. It was the early 90s. The first issue was a big industry back then, but it didn't work. I said it would be cool to solve that, and that kind of diverted the PhD and academia towards starting a company.

So I am not one of the founders of BioCatch. I knew the founders. They came to me when I was one of the founders—it gets complicated. I was one of the founders of an investment platform called OurCrowd. OurCrowd is a leading global investment platform for accredited investors. So it's going back to the community, the 8200 keeps on paying off. The gift that keeps giving. And then I kept in contact with the company. We actually put someone on the board on our behalf who then became the chairman. In 2008, he became the CEO. He asked me to join as the guy managing here, and then I replaced him three and a half years ago.

So we are in the field of security and cyber, but we're not a traditional security company. We look at the way you interact as a user with your bank account, either through a mobile device or through the desktop, through the website. By the way you move the mouse, by the way you type, by the way you hold the device, the angle, etc., we create a behavioral profile for you. That doesn't look at what you do, doesn't look at anything that you actually type, just the dynamics. And we can tell the bank whether the person on the other side of the screen is a genuine person—that it's indeed Steve and the behavior is similar to your previous sessions—or it's fraudsters that took over your account. We protect the whole lifecycle of the customers. Account takeover is what I just said, and that's when we protect against someone stealing your credentials. But we even look at new account origination. We can still, by the way you interact, think about how you type your social security number. If it's your own social security number, you'll type it continuously from long-term memory. If someone stole your credential, they would either paste that social security number or type it in a very chunky way. So those are the signals—we have thousands of those signals that we look at to give our customers, the banks, a way to protect their customers.

Basically still protecting. But if you think about the landscape of fraud as it evolves, look at the last few years. Fraudsters used to use tools—they would remote access your desktop or use malware. The defenses that the industry put in place are quite good at detecting those types of tools. It's relatively hard to do malware attacks. Basically, fraud went back to attacking the end user, the weakest link. So you see more and more scams, more social engineering. In the UK, pre-pandemic year 2020, every month there was a new scam. People calling up pretending to be the bank: 'Give me your card, you've been hacked.' In the UK in 2019, that became a pandemic. Usually, the way it goes is someone would get a call, usually a vulnerable person, an older person, and would say, 'I'm calling from Sky, the cable network, and you owe us eight pounds.' They would give their debit card, pay the eight pounds, and five minutes later they'd get another call saying, 'We're calling from Barclays.' Assuming they know who the bank account of that user is, so 'We're calling from Barclays. We just stopped a fraudulent transaction. That was for eight pounds. But because you gave your debit card and that debit card is connected to your bank account, we open a new account for you. Let us guide you through transferring all your money to that account.' Within 45 to 50 minutes, they clear the whole life savings of the person. That's not normal behavior. Someone is dictating and telling them what to do. So that's a very sophisticated and cruel con.

I think it's a unique combination. The company, on the one hand, is a big business, a real business. We're growing 50% a year. We're very targeted on the largest financial institutions. Second is that the technology, as we say, is all AI, all machine learning, deep algorithms. And the third is actually working on the good side, the good guys, protecting people from losing their money.

I think the combination of the brainpower on the computer side and the big data that can create—neural networks are not a new concept, they started in the 80s, but we didn't have enough resources or enough data back then. They are coming together. Moving forward, it will be these types of algorithms, but in conjunction with algorithms that are looking also at features. Because if you think about what we're doing, it's not enough for us to tell the bank this is risky; we need to tell them why we think this way. Neural networks are not giving good answers of why; it's more of a black box. We need to tell them we think you should not accept this credit application because the user doesn't know the data that they type in. So explainable machine learning models is where the industry will go moving forward.