When Deepfakes Steal $30M: The New Edge of Cybercrime

**Kevin Tian** (0:00)
I'd say the bleeding edge is the stuff around deepfakes, right? Because just in the past year, and I mean, this is public knowledge, we've seen eight-figure instance, right? Where people wire $30 million to the wrong account.

**Joel de la Garza** (0:14)
You know, security famously is an industry that's just cat and mouse. It's the bad people adopt a new technology, the good people adopt a new technology, and it just rinses and repeats.

**Kevin Tian** (0:24)
AI has just gotten so intelligent that it's actually gotten better at certain tasks than our original stock analyst team. And so that's certainly been a big breakthrough for Doppel, is seeing these LLM agents get smart and smart over time.

**Joel de la Garza** (0:38)
It took probably 20 years maybe to get to the end of that road and then to have these things pop up and just overnight, right? Anything that touches and makes decisions on data became irrelevant.

**Kevin Tian** (0:49)
Right.

**Joel de la Garza** (0:49)
It was just like 20 years gone. And I have to say, we look at hundreds of companies and you guys are definitely one of the ones that are at the forefront of this stuff from day one.

**SPEAKER_3** (1:00)
Today, we're talking about AI and cyber security, specifically the surge in social engineering attacks. a16z partner Joel de la Garza is joined by Kevin Tian, co-founder and CEO of Doppel Security. Kevin went from building dispatch systems and flying cars at Uber to tackling one of the fastest changing problems in security, impersonation and fraud at internet scale. They cover how generative AI is changing the threat landscape from phishing emails to deepfake scams. Why outsider founders often spot new angles in security and how Doppel is using LLMs to automate what was once a manual services heavy business. We also get into the cat and mouse cycle between attackers and defenders, the monetization of fraud, and why the biggest bottleneck now is engineering time. Let's get into it.

**Joel de la Garza** (1:48)
Hey, thank you so much, Kevin, for joining us. I know you are the CEO and co-founder of Doppel. Awesome to have you here and would love to talk a bit about what's happening with security, specifically, and how Doppel is responding to it. And then really focus in on sort of how new technologies are helping you guys build and solve very difficult problems that have historically kind of plagued your corner of the security industry for a while.

**Kevin Tian** (2:10)
Oh, for sure. Yeah. Well, first of all, thank you so much for having me, Joel. Yeah, lots of interesting stuff jumping into security. For those of you who don't know, actually, before Doppel, I was actually a software engineer at Uber, and I worked on things like dispatch systems and flying cars. So the security world has certainly been new to me, and it's been super fun just diving deep into the space, for lack of a better word. I think top of mind for me, though, and top of mind for what we live and breathe every day here at Doppel, is just how's the threat landscape changing? And AI is a huge piece of that. I know it's a big buzzword at times, and it can be marketing speak, but it's certainly been eye-opening to see how that's changing the threat landscape.

**Joel de la Garza** (2:51)
Yeah, absolutely. And it's also really interesting. I think one of the things, I think we may have talked on this in the past, one of the crazy things we've always noticed is that non-security founders, like security founders without a background in security, seem to always build some of the best security companies. I'm really curious why you think that is, as a non-security founder that's built a really cool company.

**Kevin Tian** (3:13)
Yeah. Well, it's funny because, I mean, for us, we think about, of course, what are the greatest security companies out there? And of course, you've got the model where you do have folks who've been in security for a while, right? Like George Kurtz and what he's done in the CrowdStrike and a lot of respect for what they do. And it's a company that we certainly aspire to a lot. And then we do look at companies with non-security founders, like, you know, maybe abnormal, right? As a good archetype there with what Evan and his team has done over there. I think for the latter type, for the latter archetype, what makes it so interesting and where it works is where, look, if the space is has really been struggling, for example, something around automation, if the space has really been struggling around, you know, something really hardcore technology wise, that's where I think it can help, right? The non-security founder background where you've got folks who've been incredible engineers at companies like, you know, of course, Twitter with what AbNormals team has been at or Uber, in the case of me and Rahul. And I think a lot of the problems as I've gotten into the security space have, you know, it's become clear that there's a lot of repeat problems around automation, positives versus negatives, right? That precision recall balance and those are kind of universal problems for people who are experts at machine learning and AI.