AI Leadership Insights: Financial Crime Prevention with WorkFusion’s David Caruso

Hello, and welcome to the AI Leadership Insight series. I'm Amanda Ani, and with me today I have David Caruso. He is the Vice President of Financial Crime Compliance at Work Fusion.

How are you doing? I'm doing good, Amanda. Thanks for having me.

Thank you for coming on the show. Can you share a little bit about Work Fusion and what services do you provide? Sure.

So Work Fusion is an AI agent company, and we focus solely in one space, which is financial crime compliance. Uh, so we have agents that help financial institutions of all types and sizes from the US and around the world, uh, execute a lot of the daily tasks that are required in order for them to comply with all the various laws and regulations that they have to comply with. Just one example, uh, that's been in the news a lot over the last few years, whenever you, uh, read about the US government sanctioning to a Russian oligarch or, uh, anyone like that, financial institutions have to search all of their records and determine whether or not they bank those people.

Well, that can be very onerous, very manual in nature. W among what we do is we provide AI agents that perform those tasks and do that work that allows banks to comply with those, uh, US and other, uh, rules and regulations very quickly, very efficiently, and, uh, yet, you know, much less cost. Okay.

Yeah, that's a, a big challenge, I imagine. So our topic for today is artificial intelligence and how it can be used in financial crime prevention, anti anti-money laundering and compliance. So from your experience, let's first talk about what are some of the biggest issues and concerns, um, when it comes to money laundering and compliance?

Okay, well, the one issue that has existed since any of these rules and regulations when in place, and some of them are 50 years ago, is every year there's more money laundering and there's more fraud. So it just, this sort of, the nature of human beings, there's a lot of opportunity, there's a lot of money, there's a lot of crime. And unfortunately, all of that at some point involves financial institutions and my SF financial institutions.

You can think of banks that we know, but also a lot of new company payment companies, a lot of FinTech companies. So it's sort of anything that moves money, uh, or has a customer is, is prone to this. So with this increase in crime that unfortunately continues, uh, there's a great need for a lot of people to help financial institutions either prevent this ideally, or if they can't prevent these types of customers and activity, they have to detect it and report it, uh, by law.

And so traditionally that's meant that year over year financial institutions have to hire more people, invest in more systems, and AI is impacting that because AI is now able to do much or i I say, significant amount of, uh, the work that's involved in detecting, investigating and reporting, you know, this activity. And I'm happy to talk more about that, uh, or, you know, take it any direction you'd like to go. Yeah, absolutely.

I'd love to hear some actual, um, use cases for AI in assisting. Okay, sure. Like I, I mentioned at the top of our conversation, there's a lot of work that's involved in what we, in the industry call screening.

Uh, maybe many listeners are also familiar with the concept known as KYC or know your customer. That's a concept that seems to have made it into just for the general, you know, average citizen discussion these days. And so in those, those areas of screening, uh, a lot of this work, and you can imagine, I just think about how many people there are on the planet, how many people want banking relationships or financial relationships.

When those customers, uh, sign up for accounts, they have to go through, you know, what's called an onboarding process. Well, AI agents can accelerate that. They, they automate much of it.

Let's just take for an example, uh, where a customer might snap a photo of a passport or a driver's license. Well, AI can extract all the information on that, right? It can, it can read the license, it can read the passport, it can identify the things like dates of birth, addresses, uh, names.

And so what a agents can do is they obviously can do that much better, much, much faster than a person can. And so that's just one element of, of knowing the customer. Uh, other, other use cases involve actually detecting potentially suspicious transactions.

So let's say that they, they are a customer of the bank. They've gotten through the KYC system, but they're actually, maybe they're who they say they are, but who they are is actually a bad person. The bank just doesn't know that yet.

So they begin to engage in activity that can be suspicious. For example, a lot of cash activity or maybe wire tra uh, unusual amounts of wire transfer activity to all different parts of the world that, uh, in those parts of the world might be associated or have a, you know, a history of being associated with corruption and crime. Uh, so there are analog systems that detect that.

But the problem is those analog systems, and what I mean by that is technology that's 10, 15, 20 years old, they detect a lot of that activity, but unfortunately, much of that activity isn't actually suspicious. It's, it's, it's acceptable. It's, it's non suspicious.

But banks have to hire, you know, collectively, hundreds of thousands of people are working on matters today, sorting through this, uh, these transactions and AI can just sort through those transactions, can spot patterns faster, can expand the networks of people that might be involved in that activity, uh, at a rate that human beings just simply can't. So, you know, it, it, it, it's better at detecting, uh, now what's interesting is a lot of this detection is what we in the industry call level one. So a lot of it's that sort sorting through, it's finding the things that might require more, more time, uh, more experienced people to look at.

So what we're doing is we're presenting those matters to the person, the, the trained investigator or analyst, uh, faster. And in some cases, a lot of the rudimentary document and information gathering that they would spend unfortunately hours doing well now that's in front of them so they can spend time, you know, making decisions, rendering judgment, uh, and hopefully stopping the growth of crime. So what would be considered level two or three?

And does AI have a, a use in those levels? Yeah, absolutely. Uh, and I, I've been in this field for about 30 years, financial crime compliance, but not all of it in technology, much of it on the operation side, doing the type of work that I'm describing to you.

And just as we see AI improving in every aspect of ai, whether we'll just use it for our own purposes, uh, just to, you know, plan a vacation or, or, or whatever, it's so, so that same sort of improvement we're, we're seeing similarly in our space. So to answer your question, yes, that level two work that requires more decision making, it w more of that can be done as well. Uh, so yes, because like most things, uh, well, like many things, ultimately what we're looking for in our field is patterns, is, you know, are, are, is this activity indicative of a pattern that we know to be likely to be money laundering or fraud?

And so the AI can surface those patterns, which is again, be more level two type work. Uh, and yes, you can have, you know, large language models can now draft reports. You know, the, as you might imagine, there's a lot when you're reporting activity to the government, there's a lot of requirements around what you have to write, how you have to tell a story.

And so yes, there, there's, there's that sort of level two capability is, uh, you know, so it's here already. Uh, and I would say if we had, if we spoke next year at this time, it, the rate of of improvement would be significant. It's interesting as we talk about using AI to solve these issues, I would imagine on the flip side of that, that a lot of these threat actors are using ai, um, and making all these crimes a lot easier on their end as well.

Yes. And, and one of the ways in which they do it is they create what sometimes as referred to in the industry as Frankenstein I IDs. So in other words, they, they don't create entirely fictitious people or companies.

They'll actually take stolen information or, or misappropriate information about a person or a business entity. So it's, it's true ish. 'cause there's some things in there that can be validated, but then there's other information that isn't, it's synthetic, it's, it's created.

Um, and that's able to get past a lot of sort of that onboarding review and, and screening. So, 'cause remember the, how money is moved, that that hasn't changed that much over the years. Now, it can be moved much faster, you know, almost instantaneously now, whereas in years past, it might take anywhere from three to five to 10 days for financial transactions to be settled.

So they are taking advantage of this almost instantaneous settlement, but ultimately what they want is they want that instantaneous settlement, but they also wanna mask who they really are. So you, you're sort, you're, they're, so, yes, there's, uh, and you know, as consumers, as honest law abiding consumers like you and I are, Amanda, we, we want instantaneous transactions, right? When I send my children money, they want it now.

So, so the things that we're demanding as consumers, yes, the, the, the bad guys can, can avail themselves of those same conveniences. So some of it isn't necessarily, they're creating new means and methods to, to commit fraud or launder money. They're just riding along the same rails we are.

So, and of course, financial institutions have to constantly balance the desire to give those features to customers who want them, along with the risks that those features, uh, will create. From your experience, what are some of the challenges that financial institutions face when they're implementing AI technologies? Uh, well, I think one of the challenges is simply that, and it's sort of a, a contradiction.

What I'm about to say is, is that the reason that financial institutions do this is the, the by law they have to do it. So it's, it's a highly regulated environment that in however, that, that regulation also slows things down. So, on one hand you have regulators saying, banks, you have to keep pace with this, you have to modernize.

And if that involves adopting ai, so be it. But in, before you adopt ai, you have to have very, uh, well-defined, uh, strategies for how you will do that. You have to put in a lot of governance above it.

So in a way, you have the, on the one hand saying, get at it, put in new technologies, and on the other hand is, but don't do it so fast that you do it poorly or actually increase the risk. So there, there is some, you know, that again, that's existed forever in the regulat regulated world. Uh, but I would say it's probably a little more heightened now, uh, because there's, everyone sees the rate of improvement with AI and realize that we have to adopt this.

So how do we do it, uh, wisely and safely without falling behind, you know, our peers and, and what the regulators expect from us. Yeah, absolutely. Well, AI is advancing quite rapidly.

So what do you envision as the future of AI in regard to financial institutions and crime prevention? Well, as far as financial institutions writ large, that, that's a whole nother multi-hour conversation, uh, because, you know, there's a lot of just how, how our banks and financial institutions operate. You know, many of them still operate on very outdated and antiquated systems.

So, um, the financial crime professionals within financial institutions don't, don't typically drive that modernization. So, but, but, so they'll have to follow that. But, but separately, um, yeah, there, there are sort of obviously the, the, the agent AI technologies can be adopted.

Uh, I do think it's gonna change a little bit. You know, A A ML compliance is about 20, 25 years old. There's been a lot of ways in which those operations have been developed and staffed, uh, over those 25 years.

I think AI is going to force a lot of executives and management in the space to rethink how they're staffed, uh, the, the, the, the specific skill sets that are needed. So, you know, there, there's a lot of change coming, some of it from AI itself, and then much of it from the second and third order impacts from ai. Alright, well, if there was one key takeaway you could leave our audience with today, what would that be?

Uh, that there's been, for the last five to 10 years, there's been a lot of discussion about what will happen, you know, what, how, what will happen when we see the technology, the technological change, uh, that a lot of people talk about, well, that change is here now. And you know, for someone who's been in the industry as long as I have and has heard about this for as long as I have, there does remain probably some level of cynic ci cynicism or skepticism. But that's should be in the past.

I mean, this, the things I've talked about today are happening today and financial institutions all over the world, they are deploying ai, uh, you know, all the components of ai, machine learning, natural language processing. So the takeaway should be, uh, if you're in this space and you're not availing yourself of, of ai, then you are going to be falling behind your peers. And when you're operate in a regulated environment, falling behind your peers is, is, is a bad place to be.

All right. Well, thank you so much for coming on the show and sharing your insights with us today. Thank you, Amanda.

All right. And thank you to our audience. Stay tuned.

There's more.