Listen: How machine learning reduces false positives in AML

Artificial intelligence-based security company Hawk AI today announced it has raised $10 million in series A funding. The Munich-based company is an anti-money laundering/combating the financing of terrorism (AML/CFT) platform that runs in the cloud. It will use the funds “for the international expansion of [its] business in Germany, Europe and the U.S., as well as entering new markets in the UAE and Singapore,” Hawk AI founder and CEO Tobias Schweiger told Bank Automation News . Schweiger recently spoke with BAN for this episode of “The Buzz” about the role of machine learning in automating AML/CFT efforts. Using the cloud allows Hawk AI to train its artificial intelligence (AI) with insights from multiple financial institutions, rather than relying on data from one institution to train the AI, he said. Legacy rules-based software alone isn't powerful enough to solve the problem of financial crime, said Maxime Mandin, investment director at fintech specialist BlackFin Capital Partners, which led the investment round, with participation from Picus Capital. “In an increasingly dynamic market, Hawk AI stands out as the next generation solution for transactions surveillance, with strong references not only in the mid-market but also with large-scale, complex deployments,” Mandin said in a release. “At BlackFin, we have been following the company since its inception and have always been impressed by the strong vision and the deep payment background of the founders — which is a key competitive advantage of Hawk AI.” In this podcast, Schweiger discusses the use of machine learning to eliminate false positives created by AML rules-based engines while stressing that new technologies should be seen as a supplement — not a replacement — for rules-based detection. “We believe that the balance between the two really is what it should be, certainly the next few years, simply because AI alone would not do the trick,” Schweiger said, adding that the next six months will see more financial institutions embracing new security technologies. Subscribe to The Buzz Podcast on iTunes , Spotify , or download the episode .

The following is a transcript generated by AI technology that has been lightly edited but still contains errors.

Loraine Lawson: Good day. This is Loraine Lawson, associate editor with Bank Automation News . This week on the Buzz, I spoke with Tobias Schweiger, CEO and founder of the security automation company Hawk AI. We discussed the role of artificial intelligence and machine learning in anti-money laundering efforts. So AI and ml, how did they come into play with anti-money laundering, or they used to address that?Tobias Schweiger Yeah, for us, really, I think there are two main applications here. The first one is efficiency. So the the idea of reducing the well known false alerts false positives in the industry, in a way that is risk neutral. So it doesn't avoid anything that shouldn't be avoided, so to say, but still reduces sort of waste in the process, most prominently by reducing the false positive alerts. So there's less to be worked on, if you will, the application of machine learning here is, is a somewhat traditional application of supervised machine learning for the most part. So really, learning from how the bank or the financial institutions may have decided upon some of the alerts in the past. For us, this is input, if you will, into the algorithms to start predicting what might be a false positive again, and what might be avoidable, if you will, in terms of manual investigation. So that's number one. So reducing those false alerts playing very much in an efficiency playbook, if you will. I think the other application of machine learning and artificial intelligence is of course, the more exciting one, when it's really about getting better at detecting suspicion, getting better at detecting crime. This case, then will be more of a deep learning, unsupervised type of machine learning application where it's more about finding unknown unknowns. So risks and scenarios that might not be known so far, or certainly might not have been described well enough to sort of write any rules, if you will, you know, can be detected using other sorts of applications, other technologies really, which are kind of fishing for the Unknown, Unknown, in an effort to be quicker and more precise in identifying that criminal behavior.Loraine Lawson And you think that AI and ml, are critical to that? Like, are there other technology solutions that could do the same work? Or is this something banks are gonna have to adopt if they want to comply with, say, the Bank Secrecy Act?Tobias Schweiger I think over time, that may be the case. I think there's no real technology. I think the the mean, so far, everyone's using rules. Some are using rules in clever ways, some are using rules, maybe a quite simple sort of setups, if you will. All of that is, I think not going to go away, we believe anytime soon. So I think it's there to stay because it's the way the industry so far is choosing to, you know, formulate scenarios and also, you know, communicate with one another sort of say in terms of what we are actually trying to identify and find in terms of criminal behavior here. So that's not really going to go away. So that's one technology, if you will, and the other technology will be then, like I said, machine learning and artificial intelligence, which I think is the only way to be a bit more, I guess, clever, if you will, in identifying this behavior, because the rules alone kind of wouldn't do the trick, or aren't actually doing the trick very well, at this point. So I think the industry is becoming certainly more and more aware that AI is to be used, if we were going to do that, right, if any ecommerce fraud prevention or sort of non AML type of applications, which is still similar enough to what we're trying to do here. You know, it's it's quite, it's quite easy to see that there's a lot to be learned from, let's say, ecommerce fraud prevention, which is, I think, further ahead in terms of applying those, those algorithms and those things. I think the BSA or regulation in general, I think, hasn't gone, as far as, you know, pushing for the specific application that we're talking about here. But yes, you know, I think there's a few, there's a few tendencies, I think there, those are some of the opening in that direction as we, as we speak, but it's not really in the law yet. I would say. Loraine Lawson Yeah, speaking of that, I wanted to ask you, that just passed this year, the I'm sorry, the anti money laundering act 2020, passed this year. And so what a base need to know about that, as they look forward, look forward to setting say their budgets for the next year, fiscal year, or just preparing for this year. Tobias Schweiger I think there's a few bread and butter things that I probably wouldn't, you know, talk about, here, you know, one can just read the, you know, read the sort of headlines there, I think what I probably would concentrate on in terms of what the what the act of 2020 is all talking about this, it's a very, it's a somewhat midterm, I think maybe even long term view that that is, is being taken there. I mean, you can see or read about tech symposiums or also, subcommittees really focusing on sort of innovation that are now you know, basically, you know, showing up there, let's say, as a as a to do for some of the market participants or regulatory participants. So for us, certainly as, as a company, there's a lot of, there's a lot of hope, really, that that there will be an increased dynamic in terms of looking at modern technology, looking at AI, also to some degree, and cloud and all those kinds of good buzzwords, really, you know, by all the market participants that would open the doors for applying that technology, which we really believe is going to make a drastic difference to how we run those processes today, or ideally, how we run those processes then in the future. So So that sort of door opening, I think is what you know, makes me very enthusiastic about what's happened. So to say in the, in the in the in the BSA act. I think the other things, you know, for us as a transaction monitoring and screening provider, so to say aren't necessarily directly impacting much of it, when it's about UBOs and, and reporting, and those kinds of things, because that really is a little bit the other side of the equation, if you will, but from a transaction monitoring standpoint, I think, you know, the opening the doors in terms of innovation technology is what I'm excited about. Okay, and Loraine Lawson what role does electronic ID verification play in monitoring fraud? And is that is that a form of automation at all? Tobias Schweiger Well, it's not, it's not really, I think, very close to monitoring transactions. As such, it's really more the KYC part of the equation, which is a very important one, also in terms of anti money laundering. So it's more about the process that where a new customer, let's say an individual or or even a business would be, you know, signing up with a new financial institution opening a bank account in its simplest form, maybe, right? And that's where electronic ID verification does play a significant role in sort of automating that process, automating what, you know, may have been done manually by checking passports or ID cards, you know, for the most part in the past. So, yes, it's a way of automating and it's, it's, it's certainly very good dynamic, specifically in the US, as we see around providers doing doing that kind of, you know, stuff, if you will, Loraine Lawson does that entail electronic ID verification, what does that entail? Like? Can you explained it a little bit? Yeah, I Tobias Schweiger mean, I'm, I'm not a specialist in the field, to be honest with you, but but I mean, as far as I, you know, of course, to see the what's going on is it's really about identity, firing on a account opening. Individual, let's say, by using video, using online electronic means, you know, using it Using ways of basically scanning passport or ID pictures, you know, looking closer at what video stream, you know, might be, you know might be, might be able to sort of contribute to a decision making about is the person at the other end really the person that the person says it is. So so that those kind of things, I'm not a specialist, as I say so. So certainly, certainly there will be other companies, you know, much more deep in this type of technology here. Like I said, we will be working more on the transaction side of the equation when the customer is already here. That's when we look closely. Loraine Lawson Okay, he looked very closely at that. What he's what I explained a little bit more about your company and how you automate. Tobias Schweiger So using machine learning, we talked about this a little bit previously, right, the way we automate this, basically by learning from human behavior. So just imagine people in the bank investigating suspicious cases, right, investigating the alerts, as we call them, right. And, and using that information, how the human may have behaved on specific alerts is an input into our algorithms, which then allows us to automate that behavior in a way. So next time, something similar, you know, a similar case, a similar suspicion kind of, you know, ends up on the desk, again, if you will, you know, doesn't necessarily have to be looked at, again, by human, if regulation allows that to happen, and if the bank wants it, so, and as a result of that the automation part of this would be that the case gets auto closed, or somehow, in some automated fashion worked on rather than the human actually looking at it again. So that's one way of automating, you know, basically understanding what the human would do, but then letting the machine do it. Loraine Lawson Oh, okay. So sort of, it looks at, I guess, situations where there was a false positive that way, for example, Yes, Tobias Schweiger exactly. So so looking at a false positive looking at, you know, at a case that's similar enough to something previous, and then as a result of that inferring, basically, what now has to happen, if you will, in that, that sort of action, then, you know, would be would be executed by the machine instead of the human, which saves the human work, as you can imagine, right? That's the whole purpose here. But you know, because, you know, because the underlying algorithms would be precise enough in identifying that something is very similar to something that's already been looked at, for example, it can be good enough to sort of auto close a case like that without having to, you know, bother a human again, if you will. Loraine Lawson So what do you think we'll see new in the next six months in terms of security and automation or security and attacks, or just sort of looking ahead the next six months? What do people need to have on the radar? Tobias Schweiger I think what we, what we'll see new, I think, is that, that a large number of financial institutions, also mid sized, and even small ones, will start to pilot and test and try out more modern technology for what we're doing here. So I think that I think even the BSA piece, but also in general, I think there's a there's a there's a trend, I would say, of smaller institutions, now looking at technology like ours, and in of course, there's other players in the market. And, and, you know, try out, try out machine learning, try out artificial intelligence, and maybe combine it first with the legacy system. So it's not about replacing it all together, somehow. Sometimes, it's really just, you know, combining, you know, some module of some new player, you know, with an existing legacy system, to get comfortable to, you know, start getting a bit more, you know, trusted, so to say, and really, you know, you know, develop an opinion about what new technology may be able to do in the efficiency of the process, and also in, you know, being more compliant, if you will, or, you know, running a more solid compliance process, so to say so, I think those kind of tryout type of behaviors I'm certainly seeing already today, and I think there will be an acceleration for the year with maybe larger things than going life, if you will, next year in two years after. Loraine Lawson And what do you think is driving that like when were smaller banks doing that? Now? It's just, Tobias Schweiger I think in the US specifically mean, not only there of course, you see that in there said there's a there's a consolidation happening with very small banks, midsize banks, regional banks, right. So there's, there's certainly, one trigger for for optimizing system landscape is, you know, two banks kind of coming together. So it's a choice has to be made has to be made about you know, how to combine systems, how to reap the synergies that the deal may have been promising, right. So so I think M&A or consolidation in the market, I think is one driver. Cost pressure might be another one. And thirdly, of course, it would also be a general sort of willingness to use new technology in an effort to you know, run a better and more solid compliance process. I think that will be the third third reason for me that they know that you know what I said, will come through even more. Loraine Lawson: You’ve been listening to the Buzz, a Bank Automation News podcast. Thank you for your time, and be sure to visit us at Bank automation news.com for more automation news. You can also follow us on Twitter and LinkedIn. Please don’t hesitate to rate this podcast on your podcast platform of choice.

Listen: How machine learning reduces false positives in AML

Hawk AI raises $10M in funding round

Loan process automation fintech Blend headed for IPO

Bank of Israel testing Ethereum tech in digital shekel trial

Related Posts

FIs should treat AI agents as employees, not tools

Building trust with agentic commerce imminent

Risks surface when forgotten AI agents run in the background

Bank of Israel testing Ethereum tech in digital shekel trial

EMERGING FINTECH DIRECTORY

FinAi Podcast

SPONSORED

The Next Generation of Cyberthreats: Powered by AI

Build an Antifragile Strategy to Outperform the Market

How AI and Product Experts Turn Fuzzy Requirements Into Focused Dev-ready Roadmaps

Connect

Welcome Back!

Retrieve your password

Continue Reading with FinAi News Premium - Less than $2/Day

Listen: How machine learning reduces false positives in AML

Hawk AI raises $10M in funding round

Loan process automation fintech Blend headed for IPO

Bank of Israel testing Ethereum tech in digital shekel trial

Related Posts

FinAi Podcast

Connect

Welcome Back!

Retrieve your password

Unlock This Article

Continue Reading with FinAi News Premium - Less than $2/Day