Trading Tomorrow - Navigating Trends in Capital Markets

Deepfakes, Data, and the Battle for Financial Trust

• Numerix • Season 5 • Episode 49

Share episode

The same tools fueling financial innovation are also being weaponized to exploit system vulnerabilities. From deepfakes to data breaches, regulators and institutions are racing to keep pace with a rapidly changing threat landscape.

In this episode of Trading Tomorrow -- Navigating Trends in Capital Markets, Krik Gunning, CEO, Fourthline, and a leader in compliance innovation with a background spanning banking, entrepreneurship, and fraud prevention, joins Jim Jockle to discuss how AI is reshaping trust, regulation, and security in global finance.

The conversation explores the growing sophistication of AI-driven attacks, how machine learning is strengthening defenses, and why collaboration and explainability are now central to the future of compliance and fraud prevention.

SPEAKER_01:

Welcome to Trading Tomorrow, Navigating Trends in Capital Markets, the podcast where we deep dive into technologies reshaping the world of capital markets. I'm your host, Jim Jocko, a veteran of the finance industry with a passion for the complexities of financial technologies and market trends. In each episode, we'll explore the cutting-edge trends, tools, and strategies driving today's financial landscapes and paving the way for the future. With the finance industry at a pivotal point, influenced by groundbreaking innovations, it's more crucial than ever to understand how these technological advancements interact with market dynamics.

SPEAKER_02:

The same tools fueling innovation are now being weaponized to exploit vulnerabilities across the financial system. Regulators are racing to respond to the landscape varies sharply from region to region, because banks, fintechs, and regulators all trying to keep pace. With a background spelling banking, entrepreneurship, and compliance innovation, Crick brings a unique perspective on how AI is reshaping fraud detection, regulation, and trust in the global financial system. Crick, pleasure to meet you and look forward to talking.

SPEAKER_00:

Well, thank you so much for having me. I'm really looking forward to this conversation.

SPEAKER_02:

From where you sit, how how do you think financial crime has evolved with the rise of AI?

SPEAKER_00:

Well, I think my answer would be it's the next inning in a game that we know very well. And what we've seen happening over the past eight years is fraudsters are continuously trying to out their game. So they will always look for new ways to try and breach the system. And that's the hard part of being active in this industry. You have a product that's never ever finished. And for us, it's uh it's the same thing with Gen AI. Uh, I must say that the the one thing that really changes uh the game for the fraudsters this time around is the scalability of the attacks that they can launch.

SPEAKER_02:

So, what kinds of fraud or abuse are you seeing most often in this evolving environment?

SPEAKER_00:

I would say the biggest change we've seen is in deep fakes and specifically in the quality of the deep fakes. Uh, we started seeing them probably two and a half years ago. Uh, and Jim and honesty, the first ones were laughable. Uh, the team would show them to me, and you could still see from a mile off that it was a a deepfake, especially around the eyes, it was really, really clear. Um, and I think what we've seen in the in the past 12 to 18 months is a a stellar rise in the quality of the deepfakes. I think that's the biggest difference. Um, I must say on the document side, it's a bit of a different story. And I think on the document side, there's also um a lot of ways to prevent Gen AI from even coming in. But I think on the biometric side, it's it's clearly a different story.

SPEAKER_02:

Just uh last week we saw the uh the rollout of Sora 2 and uh some of the creativity that's blowing up Instagram and and other areas, it's it's incredible. Um, obviously uh certain things are very, very fake, but uh gosh darn it, they can be pretty funny too.

SPEAKER_00:

It's a super powerful technology, and uh I I would prefer people to uh make videos, videos that you can laugh at rather than trying to uh gain access to the financial system.

SPEAKER_02:

So many firms are turning to AI to fight AI. Uh, how realistic is that? And where does that work best?

SPEAKER_00:

So, I mean, I think first and foremost, it's an incredibly powerful technology, not just on the offensive side, but also very much so on the defensive side. And I think maybe one thing to clarify is is that Ford line we started investing in AI probably around eight years ago. Um, but that's very much on the machine learning side of things. And uh the good news is that there we can actually prove based on a ton of data that AI is super effective on the on the defensive side. Um, I do think uh one important point to make here is there is no silver buddha. There will not be a single check that allows you to prevent all fraud from coming in. Um so the approach we've taken instead is a layered approach where we run seven several different AI models that all look at a slightly different aspect of a of a case. And because of that layered approach, you become really effective at stopping fraud. Um, but it's something where we can really show that it delivers value today. So I think there's been a lot of discussion in the news in recent weeks and months around the future value that AI will bring and how much jobs it will uh eliminate, how much revenue uh the likes of open AI will make. But I prefer to look at the tangible impact that we can make today. And I think in the fight against financial crime, that impact is very tangible and very positive.

SPEAKER_02:

So regulation, obviously, you know, the the hand that forces the market, but it looks different in every region, every country. You know, how much does that shape what technologies banks can actually use?

SPEAKER_00:

Well, I mean I would say it's it's not so much what technologies they can use, but more how they can leverage these technologies. And I think that very much depends per region. Um, I think a great example would be if you think about the fight against financial crime, this all evolves around identity and some of the most sensitive data uh out there. Um and we're headquartered in Europe, and as you know, in Europe we have privacy regulation, the GDPR. Um and as a company, as a European company, that means that from day one, we need to build a privacy-first platform. And that is hard to do at first, but if you do it properly at setup, both from a technology perspective, from an operational perspective, from a knowledge perspective, it's difficult but doable. Um, and in hindsight, I think it it has given us um an advantage also from an international perspective, because we can see uh in the states that even though there's no federal privacy law, we've seen states like um California and Illinois adopt uh privacy legislation that's very similar to what we are used to here in in Europe. And reverse engineering a platform afterwards to comply with strict privacy regulations is much harder than if you've built it in from day one.

SPEAKER_02:

We're now several years into GDPR. How much of that has had a positive impact? I mean, obviously, it really forces a lot of companies, because of those steep fines, to take data security, data privacy very, very seriously. But you know, do you feel now that we're kind of a couple of years into this? Obviously, we've seen emergence in California. Germany is very much let's keep uh every everything in country. Uh Japan has uh high watermarks as well. Um, do you see those regulations having a greater impact or individuals and companies you know still just kind of frivolous with with their with their with their data?

SPEAKER_00:

I think it's most interesting to look at it through the lens of a consumer. And I think what we've seen happen over time is initially, I think people didn't care that much about GDPR, but over time something really interesting happened, and that's that it creates an awareness of um the value and the sensitivity of your own data. And that's something that we now see across all the markets that we operate in, that there's a very high awareness of individuals that you shouldn't share your most sensitive data with anyone at any point in time. And I think that's actually the biggest win of privacy regulation, that people are aware of that, and that will also pave the way for some of the super interesting petables we now see around identity wallet wallets where you actually will empower citizens to share information on a need-to-know basis rather than massively oversharing information with institutions that actually don't require all that information.

SPEAKER_02:

Well, when I go to a cocktop party, I always ask people what's their grandmother's maiden name, what's uh the name of their first pep. No, just joking.

SPEAKER_00:

I mean, I I think it's actually a really fair point, but I mean, I think if you if you want to check whether someone is of uh legal drinking age, the way that that question typically gets answered is by spitting all the beans, right? Your nationality, date of birth, place of birth, full legal name, address. People don't actually need to know that. They just need to know whether they they can serve you alcohol or not. And I think it is an example where there's now a bigger awareness of people to not just share everything with everyone.

SPEAKER_02:

So confiance and innovation, right? The two opposing forces, right? They'd be a pull in opposite directions. I mean, how do you see institutions managing this balance?

SPEAKER_00:

I think, Jim, my answer would be if I look at it from a personal perspective. Um, the first really big wow moment I had with AI was back in 2017. Um, Netflix aired a documentary called Alpha Go, which tracks uh Google DeepMind's efforts in in training AI first on chess, but then later on the game of Go. And ultimately uh you see during the documentary that they beat the world champion. And for me, that is both an example of how powerful the technology is, but more importantly, why it actually works exceptionally well in a highly regulated environment. Because if you ask 100 companies what they think of regulation, 99 will tell you that it prevents them from doing business. And I think I'm beyond one app because I think for me it actually sets the rules of the game. And if I know the rules of the game, that means I can find a path to victory. And that's what that documentary showed. And that's also the way um I look at regulation today. As long as you give me clear regulations, we will apply technology to come up with the very best solution for both the financial institutions as well as for their customers.

SPEAKER_02:

So collaboration is essential for our prevention, whether it's as an individual and the relationship with a particular company and and just being mindful or an individual providing information and data. But what is the collaboration between institutions and regulators, especially around data security?

SPEAKER_00:

Yeah, I mean, I think the the analogy is the same one that you just gave with with the cocktail party, which is it's not so much about exchanging data and sharing data between financial institutions on a very granular and very personal level. Um, what is essential is that you actually take a bird's eye view of what's happening in the market. Um, and I'll give you an example. I was speaking to a large traditional bank in Europe the other day, and they explained the way they are set up, um, which is they have one team running KYC with one data set, they have a second team running Frog with a second data set, they have a third team running sanctioned screening with a third data set. So even within one bank, they cannot leverage a single data set. And I think what needs to happen today, and that's something that Fort Line is already doing, is you need to be able to track trends on a cross-border, cross-parton level through time. And that's where the real power comes in, because it means um you're able to track new fraud trends earlier. And you asked about the device of Gen AI. It's it's a great example where you want to know very early on what is happening in the market. And a bank alone can simply not do that. So you need to have this bigger perspective. And the advantage of the bigger perspective is that you will see more, and if you see more, you will know more, and if you know more, you will see more. So there's a network effect in there, and I think that's really where regulators at financial institutions can come together.

SPEAKER_02:

In that bank example, was that the bank's decision to ring fence the data, or was that a regulatory uh drop driven in terms of keeping those data sets or I the answer would be neither.

SPEAKER_00:

It was just they were being forced by legacy systems. So it was a bank that was a result of multiple uh mergers and acquisitions, as a result, they had uh what well is commonly referred to as like a spaghetti of IT systems, as a result of which they could not effectively use the data that they have in Germany.

SPEAKER_02:

And would the end result, because one would argue the larger the data set, the more information, uh the better the outputs uh and formats, especially in areas of financial crime. So would the would the argument be to bring all that data together through some sort of transformation effort? 100%.

SPEAKER_00:

I think I'll give you a very practical example. Um what we've done at Fortline is we have combined all the checks on a single proprietary tech stack that uses a uniform data model across each step of the process. And one of the big breakthroughs that we saw was specifically around sanctioned screening, where historically uh people are looking at just a full legal name and then they apply some fuzzy logic, and then they tell a bank, unfortunately, 8% of your clients are a potential hit. Now you go and manually investigate which ones are true hits and which ones are false positives. Uh the disadvantage of that is that it costs a lot of time and a lot of money. And what we've seen is if you actually add a couple of data points to that analysis, uh, which you know from the ID document. So let's say you have full legal name, date of birth, place of birth, nationality, a couple of other data points, that becomes incredibly powerful in making sure you never miss a true hit, but also can confidently eliminate false positives. And I think that's an example where reusing data points that you have from another process can be incredibly powerful.

SPEAKER_02:

AI brings so much opportunity. Uh everybody I know at this point, you know, the the barrier to adoption is just is lowering dramatically. But with every opportunity, there's risk. And what would you say the biggest challenges that you see in deploying it sponsorably, especially around compliance?

SPEAKER_00:

I would say it's extremely important to make sure that you deploy technology that is leveraged in a way that you can explain afterwards, and especially that you can explain to regulators. Um, and we spend a ton of time speaking to financial regulators, and surprisingly, they're much more open to AI than a lot of people realize. They're also a lot smarter about AI than a lot of people realize, and they will ask tough questions. They will ask questions around uh how the model was built, how it was trained, how it was validated, how it was backtested, but especially whether you can actually explain the outcome. And I would say that is the biggest risk. So if you deploy AI that's super smart, that will give you a result, but no one actually knows what's happening what happened. So it's like a black box, both to a financial institution, but especially towards the regulator, that's going to get shut down. So you need to be able to explain actually what happened.

SPEAKER_02:

One guest we had on, which was interesting, uh, we were talking about AI and trading, and he's giving an example of sometimes having the AI explain itself is actually detrimental to the performance, uh, because it's making potentially steps that would be illogical uh to a to a human. Uh, have you heard elements of that before? Or or was that just one person's opinion?

SPEAKER_00:

I mean, I I uh I think in general there is something that can happen, but I think specifically when it comes to compliance, you better make sure that you have an audit through and that that you can explain each of the steps in the process. Uh, because this is a super sensitive topic, right? You're deciding uh on how you can preserve the integrity of the financial system by making sure that you provide access to bodified clients and and block the ones that are bad actors. And if you don't know how we you reach that decision, um you're gonna end up in court very, very quickly. And it will very be very easy for someone who was rejected uh to appeal against that rejection if you come up with a model that spit out an answer, but no one really knows why. So I think it's it's really fundamental in the way a regulator expects financial institutions to operate to have that full audit trail and to be able to explain every decision that you make.

SPEAKER_02:

So you're uh Burke, you have a background in banking and finished. What differences stand out between how traditional institutions and neobanks are tackling financial price?

SPEAKER_00:

I think it's a it's a great question. I think, in honesty, um I I worked in banking at the start of the century, and and KOC was not exactly uh top of mind back then for uh the boards at banks, and it very much is so today. Uh and we had the privilege of of supplying our technology both to traditional banks and to fintechs. And there's a bit of cross-fertilization there because banks love the fact that we work with fintechs because they know that fintechs are obsessed about UX and conversion. But the other way around, fintechs actually love the fact that we work with traditional banks because they know they're obsessed about uh security and compliance. And I think that's really the example um that that powers the future because it means you need to be able to deliver both. And if you cannot deliver both, you're gonna be in trouble in the future.

SPEAKER_02:

So let's talk about the future. How do you expect AI and regulation to shape the next generation of financial crime defense?

SPEAKER_00:

I mean, I think it's gonna strengthen a trend that we already saw before the rise of AI, which is historically speaking, um, KOC and compliance was was almost like a snapshot. So you would say this is Jim on the the 15th of October 2025. Now we're gonna run some checks, then we're gonna file the information never to be looked at it again. So really a snapshot in time. And I think there is a a grown pressure, a grown realization with financial institutions to move from a snapshot to a movie, which means the the onboarding and the initial account opening is a starting point, but that actually creates um yeah what we call a ground truth, and that will form the basis for checks you run in the future, where it's becoming super important to make sure that the person using the account is actually the same as the person owning the account. Uh, and if you think about the risks uh with AI, it's not just about someone opening an account, leveraging AI, but much more so leveraging AI to get access to a legit customer's existing account, so via the account takeover. And I think that's where um some financial institutions still have a long way to go. So I think treating it as a one-off event is not gonna make you feature-proof.

SPEAKER_02:

So, Kirk, unfortunately, we've made it to the final question of the podcast. We call it the trend drop. It's like a desert island question. And if you could only watch or track one trend in financial tech over the next few years, what would that be?

SPEAKER_00:

So I think everyone would expect me to say AI. And I'm gonna give a slightly more nuanced answer. I would say machine learning, because I think what we've seen uh in the past two years is everyone refers to AI, what they actually mean gen AI. So I think because of the fact that Gen AI has gone mainstream, we've seen uh the adoption also at a consumer level level skyrocket. A lot of people are now using the term AI to actually describe gen AI. And the sad victim of that, in my view, is is machine learning, which is becoming a little bit like the orphan of AI when all the attention goes to the sibling gen AI that everyone loves to talk about. Whereas getting back to what I said before, if you look at the incredible impact that machine learning can have within financial institutions by both increasing the quality, massively reducing the cost, and massively speeding up processes, I would say watch out for machine learning. It's maybe undervated today, but it will deliver in the future.

SPEAKER_02:

Kirk, I want to thank you so much for your insights and uh stay safe out there. Thanks so much, Jim. It was an absolute pleasure to be with you.