(upbeat music) Hello and welcome to the Money Pop Podcast. I'm Charlene Tom, my co-host is Mickey Tesswey and we're recording live in Vegas at Money 2020 USA on a buzzing show floor where the best and brightest fintech and finance come together and Money Does Business. And here on the Money Pop, we're joined by some of those very same minds to discuss the latest things happening in the industry. What they're excited about and the future of money. Today, we're joined by Ricardo Amper, who's the founder and CEO of Incode Technologies and we'll be discussing fraud. Yes, we will. Nicely done, Charlene. Ricardo, thanks so much for sitting down with us. I thank you so much for having me excited to talk about fraud and all the accounts with it. Amazing. You go, he's done. Okay, to kick things off. Ricardo, can you start by telling us why the conversation about AI and fraud is so critical right now? What are you seeing in terms of these threats? There are many things that are happening with fraud and AI. They are subtle categories. First category, volume is exploding. The cost of doing fraud with AI, which in AI tools like Deepfakes is almost zero. Anybody can do a deep fake, take sort of two. You can do a great deep fake. You take Gemini 2.15, you can do a perfect deep fake of an ID, meaning a digital version of whatever driver license you can. At the same time, we're seeing not just the volume explode, but also the velocity of the speed. Every time there's a new tool, there's very strong fraud rings out there that are used into tools creating their own. And so we're seeing an explosion of fraud where most of the fraud that we're seeing is now generated with AI. It's impossible for humans to look and the speed is probably about 10 times faster than we used to. And so with that, you have to ask, okay, what you can do with it. And it's really AI fighting against AI. It's how do you build these tools, how do we come together as an industry and a society to be able to tackle this problem? The point about Deepfakes is so fascinating because I'm not gonna lie to you. Some days after a long day at work, I like to go on Instagram. I do what the kids call Brain Rock. And I'm just there scrolling away. And you know what I've become addicted to recently is these videos of babies doing podcasts. It's like these videos of like a random podcast, but they put babies on it and it looks so realistic. And it's like, I'm like, wow. And this is people who are trying to create something for fun. They're not even incentivized to make money from it. That just, to me, is like the best way to understand just like the pace at which things could be replicated nowadays. And going back even a couple of years, I remember earlier at the start of the war in Europe, you know, Ukraine, Russia, I remember. There was a few actual government ministers. I think there was a in Germany that got deep faked by like, you know, videos of I think it was like the mayor of St. Petersburg or something like this, right? But like actually like real government ministers were falling for this. So it really demonstrates the kind of size and scale of the issue we're looking at. In terms of from your perspective, when you're thinking about businesses and the businesses you help as well, and when they're creating identity verification solutions for onboarding customers and so forth, they're facing real tension right between security and conversion. You know, we always talk about specialty and payments to your right about seamness, payments, reducing drop-offs, and of course, of securities to tie increases the chances of drop-offs due to friction. And then of course, the other side is of us to easy, then it just becomes an entryway for a huge last-floor to occur. Talk to us a little bit more about this dilemma in detail from your vantage point and really how it plays out. Right. Everybody thinks that the price of security is friction. I just think friction is a result of bad design, bad strategy. So for us, if you talk about what are the things that we're doing to find, for example, those deep fakes, we have a product that we announced recently called DeepSight. And what it does is protect your device, protects the looks at the camera, looks at why you're putting there, it protects the connections, again, deep fakes, understands how you are interacting with the phone. And with this product, we can significantly produce the amount of frother without even creating friction. Actually, there's a new paper by Purdue University that came up last week that says that ink odys the best deep fakes company in the world of any deep fakes in terms of detecting deep fakes. Not creating deep fakes. Well, actually, fun enough, in order to become the best deep fakes detection company, you have to be one of the best deep fakes creation companies. So you did, you had to go work the journey. Have a, we have a, a fraud, a lap that has about 200 tools, some of the ones that we create, where you have to think about new ways of having real looking babies doing podcasts to be able to train these models. And so what this paper says is not only in code is the best at the technical defects and the lowest false positive, but actually we are the best when you're looking at the camera. So that's one of the thing. But the other thing that you have to think about friction and security is that we cannot fight it alone. We cannot because AI is going to defeat every security, protocol out there that you can think of. But what it won't defeat is the system as a whole. The system as a whole and all us together feeding data in a privacy centric way is important because, you know, 20 years ago, nobody cares much about privacy and privacy become really important. I think it's important to human right. And there is notion that with privacy, you really cannot create great security. But right now there are tools with zero knowledge that you can ping a graph, get this, this patterns that everybody's exposing their data without sharing the data. And you can go and find fraud at a way that you can. So talking about friction when, when system is looking at each thing as a signal, not as a wall, but as a signal, we can create a flow to open a bank account or approve a payment that is invisible. Our goal is to create what we call invisible security. It's there. It's by the way orchestrated by an AI. It's like thought and orchestrated by AI and it's fighting other AI, which is, I think, really interesting. And I love it before, I let Sean Lee know how to connect. I love that point around the data sharing. I think data sharing sounds like, in previous years, you like, sold off in world gardens were an important way to protect data leak and making sure that you had control over the data that you've taken. But nowadays it seems like actually data sharing is the ultimate way in which as an industry, collectively, we can fight against fraud, right? Because often these forces would have gone through one of these networks. They would have already probably been onboarded checked at some different place. And I assume a lot of the times, a lot of these forces occur because there are gaps in between the way in which different companies and different players in the ecosystem come to say with each other, right? Exactly right. It's a design flaw. So if everybody is wall-off, you see all these bad people, which is a very small percentage of population doing bad things at bank A and then kind of proving and trying to look at every door, see which are weakest and doing bank B, B and C. But if you subscribe to the notion with AI agents, imagine agents going and doing fraud and impersonating you in all the different banks in real time. If we can't, that's your identity is being used by an agent, a rogue agent. We should be able to tell the rest, right? In a way that the seal privacy is centric and it's possible now. And so that's where we're encoders is obsessed about. Is how do we create a better system overall? Not just great AI, yes, but an overall system where everybody's participating and there we can defeat. I love the way that you said it's a design through a couple of times. You speak like a real engineer, you know? Amazing. I think that point on collaboration was brought up a few times at the Ford Summit as well. So it definitely rings true. You mentioned DeepSight, one of your tools. And we hear a lot about deep fakes and AI generated fraud, fraud GPT, you know, it's a huge tool that people are using. How accessible are these tools for commoners today? And what kind of scale are we talking about? The scale, I mean, accessible, it's almost free. It's almost free. And there's very, very strong fraud rings out there that are sharing all these tools, creating all these tools. The scale is at a scale that we have in them and it's difficult to think of because the assumption that we've worked on is that you need humans or human written scripts on computers doing bad things. We're about to start the agenteic world, right? Agents doing great things for all of us, but they're also going to be deployed millions of billions of them trying to the fraud. And so the scale of the fraud that we're going to see is for the next five years is going to be crazy. And for all these general AI revolution to work, it has to be trustworthy. It has to be a lot more good than bad. That's why here on the on the on the Monipot stage, we launched a right, agenteic identity product, which is how within your change of PT, your cloud, name your tool, you can go and open bank accounts, move money, all of that while protecting. And so for us, it's incredible. I think for the whole industry is incredible time because it is a very strong tailwind, but I think the magnitude and the speed is something most people don't realize. And we're very focused particularly on the speed side. It makes a lot of sense. You touched on the agent stuff on agent AI, which I. So exciting, right? In terms of, you know, the promise is that I'm very bad at booking stuff. Whenever we have to come to money 2020, the team have to bug me over and over again. And now I'm hearing the idea that there's going to be an agent, Mickey, more organized, Mickey going to go book my flights for me, everything which sounds amazing. But as you know, better than anyone, identity verification, even today without agents is not something we've sold for completely, right? Like, of course, we made huge amounts of progress with the likes of, you know, with you guys with so many companies, you were doing great stuff to make sure that customer onboarding is without as much friction as it used to be even five to 10 years ago and as as secure as possible. As you're thinking about, you know, the future of this, this is this agentic future where we're all going to have thousands and millions of agents doing things for us. What are some of the key things we need to resolve, even that we've yet to fully figure out with the identity verification to unlock the real opportunities of AI rather than to, you know, walk into the kind of potential risk that exists as a result of it. Yeah. But whenever, when we talk about, you know, catching a roster, every time we catch a roster, you're actually helping hundreds or thousands of users grow and do more things, right? We are focused on the, our mission is to power world of trust, right? Because we're not about frauds about trust, how do you create that trust? Right? How do you do more things? So I think there's a few fundamental things. Number one is you want to do an onboarding flow that's intelligent. And so when I see, when I say that everybody uses trust graph, which is the product that helps do these zero knowledge, you know, sharing with that product, what it does, you create a much and more intelligent flow. So instead of just putting big walls in front of you and see who I can get, you know, it's like the bazooka. This is what happens. Now that's why it's so more friction. You're having a sniper based approach where we really use that information to make sure that we apply the appropriate level of risk. Right? That's number one. The other two, the other, the second one is really interesting. What, well, we could see on the LLMs is that you generalize, right? The very quickly, there, there can be any model because it uses feedback that you train them for to do one thing and it ends up doing five things. And so these year at Incode, we launched the first foundation, AI foundation models for fraud and for trust. And while we saw what's super interesting. Number one is let's see, let's say that there's new Gemini coming up and you can do all these types of IDs. It will take normally probably like six months, four months to have a hundred thousand samples of something good and bad, you know, the hot dog, not hot dog thing and be able to go train a model, deploy a model that takes months with this foundational models that we created. It takes a day or two days and maybe ten samples only. So that means you can go much faster, much faster. The second thing is that the one you train them to do one thing, it ends up learning with all the data, all the collaboration and so learning five things. And so if you take that into the future, you can imagine AI, not just AI fighting against AI, but AI being autonomously finding solutions for the new problem. So at Incode, that's that's what we're building. We're building an autonomous system that fives fraud. And for that, you need great engineers and a lot of customers we have, most of the big banks in the industry and in the US, slammer on the world, our customers of us. And this concept of zero knowledge sharing together will take us there. A quick question, just a quick follow up on just because you specifically just pointed out, you know, your customers, you can't some of the biggest banks in the world and financial institutions as customers. And what we're seeing in financial services right now is quite an interesting tension, right? There's the tension between, so we know AI is incredibly transformative. And the tension seems to be on the one hand is how do you go to market quickly and take advantage of this momentum? And then on the flip side of it, it's actually you need to move a bit more slower because some of the guard roles don't already exist, right? Probably that's the biggest tension for most executives who are looking at the world today, right? If you're trying to move a bit slower and be responsible, and then you're reading the Wall Street Times or whatever, Wall Street Journal, and wherever you read your business, you're seeing all of your competitors probably launching an AI thing and you're thinking on my stock prices sliding, the board is going to come for me. So there is a very tough tension for leaders to have the face. And as you're kind of, you know, talking to your clients and listening to their concerns, what are the kind of fundamental things that are telling you with respect to AI, with respect to fraud and AI agents and how they're struggling to really make a difference. Yeah, it is very interesting, as you say, you have the Trump administration on one hand being incredibly supportive of AI. On the other hand, saying, you know, drug trafficking is terrorism and banks being shut down with an executive order around the world. And so you can see the tension. If there's a few transactions that you didn't know of, they can shut down your bank. But at the same time, you're seeing all these fintechs and all the pressure between the bank. Right. So what banks are telling us are number one, the conversations of trust shouldn't be at the fraud, like VP of fraud C-Sol level. It should be a board discussion, though, because it is reputational risk. If companies are not protected against the defects, another example is the you're hiring people. You're my hiring just someone who was not who they say they are and that can be a reputation. So it has to be at the board level and be because when security incidents happen, the companies tend to keep it to themselves. It takes more time. There's a lack. So you see all this progress and we as in could see all these bad things happening, but it's much worse than people think just because they don't share companies don't share it's not in their interest. And so what we're seeing is C-Sol is telling us be there, be out there, be a thought leader, educate the market on what's happening there so that we can go and adapt. And I think to your initial question, there's no choice. Right now seems like a choice. But in a year or two, the biggest companies, the companies that adapt AI, the first are going to grow very fast and the ones that don't, not only they're not going to grow, they're going to be defeated by fraud. And so we're going to see all these be shuffling and venture companies where the ones that are the yearly adopters are probably going to have a much bigger growth on the rest. I think I thought that point you just made around how companies react to fraud happening at scale to them is so interesting to me because every incentive exists for you as a big company. If you've got a big day to keep that as small a news as possible, right? Because we do know the reputation or risk of having fraud happening to at scale. If you're a big company and we've seen so many instances and especially in the US where you get big companies, you get millions of records leaked and it feels like it's regular business nowadays, right? And it sounds to me actually that is a big part of the culture that has to change with respect to the point you're making about data sharing as being one of the most impactful things we have outside of the technology piece to defeat fraud in a meaningful way, right? So I think it sounds to me like a big part of the success going forward and further kind of forward thinking businesses is actually being in a mindset to shift some of the culture around how they treat fraud because ultimately that's what is the end worry we're talking about, right? Is if you don't get on top of what's going on with AI right now, the scale of this fraudulent activity is going to be so big that reputation risk is going to be X 50X whatever it might be. Yes, and he has an impact on society as well because typically when when there's mistrust, the ones that suffer the most are the ones that have the least the ones that don't have a strong credit history or a lot of activity under credit bureaus are the ones that whenever there's no information, we build walls and they cannot access these products and services, right? Yeah. And so I think the way Incode sees his mission is yes, we have to help these companies and we have to enable governments and private companies to share in a way that's private, but at the same time, we have to take a deep focus on how we enable people who have the least, which is an opportunity with AI to to put them, you kind of rebalance society, but only if we're able to create trust. And so I as a founder and CEO with my colleagues, I feel incredibly lucky that we work on financial inclusion. We work on human trafficking when it comes to adult entertainment, how you protect communities, you make sure gambling, for example, the whole community is safe for in terms of other age kids, national security. So to me, this concept of trust that enables growth is deeply person. And that's the flip side of it too. When done correctly, for instance, right now today, we know that trafficking, no terrorism, funding and even on, lots of platforms around the world when it comes to underage content being accessible. There's huge amounts of costs involved and AI and these, you know, identity verification technologies actually have huge meaningful impacts to people's lives. So I actually really love that you kind of brought that to the forefront because ultimately that is what we're trying to do. We're always trying to make sure that people as a society are moving forward. Exactly. And what we see at Incode because we have the numbers is the percentage of people doing bad. It's solo. But then we see how many people were stopping when they're trying to open a bank account. Or we see some people that are doing human trafficking. And then they go live their lives as normal citizens, banking with their organizations, which I don't think it's, it is, it is, it is right. So I think there is an opportunity with AI to create a more accountable, you know, society that puts walls and blocks, users who are definitely
trying to the father system, but that will create a lot of economic and social prosperity, which is something I get particularly in the financial inclusion side. I get very, very passionate about it. Amazing. So much to unpack in there. We're going to have to bring things to an end for this session. Ricardo, it's been amazing to have you with us today. Again, really appreciate you taking the time to join us. To all our listeners, thank you for caring so much about the industry, about fraud, AI, and trust, as we do. You can subscribe to the Money Pop wherever you get your podcasts or find us at money2020.com. If you want to get in touch, you can find us at
[email protected].
PodcastTranscript AI