Listen: Nice Actimize launches AI-backed tool to thwart false positives
Anti-money laundering and fraud solution provider Nice Actimize announced a new artificial intelligence-driven tool Thursday that it claims will reduce false positives when conducting anti-money laundering (AML) screening on potential customers.
False positives account for roughly 99% of alerts when screening against a sanctions list for AML, said Adam McLaughlin, global head for strategy and marketing at Nice. The new tool, WL-X, uses both AI and multiple data sets to reduce that number, although the company could not specify by how much WL-X reduced false positives, saying “the results ultimately depend [on] the firm’s holistic approach to AML.”
WL-X uses both open and premium data sources, and applies AI-based analysis and biometrics to those data sets to reduce the false positive rate, which translates to less manual work for bank employees.
“It gives the regulated organization the power to screen across multiple data lists, both structured and unstructured, which gives them a lot more control over managing their risk as an organization,” McLaughlin said.
Often AI is applied to Know Your Customer procedures and AML, but where Nice thinks it’s distinguishing itself is in applying AI to both structured and unstructured data.
In this episode of The Buzz, McLaughlin talks through the back-end plumbing of this new solution.
Shares of Nice [Nasdaq: NICE] were trading at $277.63 as of 4:46 p.m., up 1.27% from market open. Nice has a market capitalization of $17.4 billion.
Bank Automation Ignite, on April 13-14, is the event for inspiring automation initiatives and investment in financial services. At the virtual event, financial services professionals can discover new use cases and technologies that are accelerating automation in banking. Learn more and register at www.BankAutomationIgnite.com.
The following is a transcript generated by AI technology that has been lightly edited but still contains errors.
Adam McLaughlin
Sure. So it’s, like, like you said, it’s a breakthrough in screening. And it’s a new solution, we launched a and the solution ultimately allows organizations to manage their screening risk, globally and across the whole organization. So what it allows organizations to do is to take in multiple data sources. So historically, and legacy screening solutions will often just look at one data source. So one list source, which looks at PEPS, which is politically exposed persons, and sanctions risk. So this is risks where countries, governments will put sanctions and restrictions on either countries or individuals or corporates, stating that these entities, individuals or countries cannot deal in US dollars or, or euros or other things. So you know, that this is a government-dictated list saying that banks cannot deal with these individuals, people. So legacy systems very much deal with historic lists, and it’s very structured in terms of the data they get, that creates inherent risk that certain data might be missed. And it doesn’t expand into the scope of things like adverse media, it doesn’t often expand into things like ships, some ships can be sanctioned, because they can come from sanctioned country, for example, like Syria, or Yemen. And so there’s a lot of risks that go way beyond just looking at politically exposed people, and individual and corporate sanctions. And we’ve broken down those barriers. So that risk by integrating multiple data sources, both in open source and premium data sources, into our streaming solution. So it gives the regulated organization, the power to screen across multiple data lists, both structured and unstructured, which gives them a lot more control over managing their risk as an organization.
Loraine Lawson
You’re using biometrics as well, in the solution and AI.Adam McLaughlin
Absolutely. So I think historically, with, again, you know, looking at Legacy screen solutions, and one thing I’d like to sort of highlight is transaction monitoring, and KYC procedures, you know, a part of the AML program, both systems have have already come into the world of AI. So they’re already utilizing AI and analytics to help enhance their detection, to help make sure that what they’re detecting is accurate and precise. But screening is often lag behind those screening solutions, Lexi have often not integrated this new technology to help enhance their screening, sort of accuracy. And so what we’ve done, we’ve been fused within the course of the platform, why we’ve infused AI, to help make sure that what’s been screened and the data that’s coming into the system, we’re able to accurately assess the right people, we’re able to make sure that the people we’re screening against, and the hits we’re getting are the same person. However, there is still a limit. So AI can do a lot and it will massively increase the accuracy. But sometimes you American text based screening, they’re still you know, there’s common names, like for example, across Europe, your Smith, for example, as a surname. So some common names, it’s still quite common in in various jurisdictions. And so you might get a very a lot of hits, coming back after disposition, those hits and say, which of these hits is the right person and actually, the person I’m screening is this a person related to these visa hits I’m getting from the list data. So biometrics takes it to that next level. So if you have a passport or a driving license or some photographic information of that customer that you’re screening against, You can feed that into the system. And it will it will check against a list of photographs. And these photographs are photographs of political people. So they might either be direct politicians, or they might be family members, politicians, they might be sanctioned individuals as individuals who have got a sanction against them, or they might be terrorists. So, you know, the solution has taken videos, news articles, and open source photographs of individuals into a database, so we can screen against that database. And if the photograph of the customer matches this database, it will return an accuracy score. And so what that can do is take the alphanumeric hit that you’ve got, and give very good precision, that the person you’re looking at is exactly the right person, because you’re actually going to photograph as well as alphanumericLoraine Lawson
text. Okay. And when it says financial services organizations use this, but when you see banks using it, what what would be their use case for this? I use a screening, but like in which situations?Adam McLaughlin
Sure, so screening is absolutely foundation of, of AML. And it’s the foundation of compliance. So banks and financial institutions have to do this in the regulated sector. So it’s not just banks, you know, insurance companies, and any organization that is regulated, have to do this activity. So it isn’t just the banks, you know, anything that deals in finance, and has to be regulated by fincen. And other regulatory bodies around the world, have to do screening. And so this will happen right at the beginning of a relationship with a customer. So before you even take them on as a customer, you have to make sure that they’re not sanctioned, because if they’re sanctioned, you cannot be dealing with that customer. So that is the end of the relationship.Loraine Lawson
To this network, we primarily write about banks. So how did they do it now.Adam McLaughlin
So at the moment that they do use wind solutions at the moment, but what they haven’t, what they don’t do is they don’t infuse AI. And very often, they just use singular lists. So they will use list providers. And there’s a number of them out there. But often banks will plug into one list provider to do their checking. And we’ve got one list and you don’t have the AI that sits behind it. What that generates is a lot of what we call false positives, that’s a lot of noise, if you like. So the sum of some figures we’ve gotten, there’s a report from eBay. And it says roughly that the number of false positives is about 99%. So the results, I get back, roughly 99% those results, I get back from systems without using AI to try and clean that data. You know, 99% of these hits are actually false, which creates a lot of work a lot of effort, and ultimately cost a lot of money for the banks to manage their screening programs, because they have to employ people to look at the results to reduce those 99 or workflows, 99% of false positives. So that’s how it works the moment and what we’ve done is we’ve taken change the curve, if you like and said actually, by utilizing AI, by utilizing biometrics, by ingesting additional data sources across that lifecycle, we can actually start enhancing that you accuracy and precision on making sure that the person you, your customer, your prospect is the right person who’s who matched against on a peps list or sanctions list or an adverse MEDIA list. And I just like to sort of go back to point so you do onboarding. So you have to make sure that from day one, you understand that the risk of that customer but it doesn’t stop there. You have to screen throughout the lifecycle because everyone changes. So for example, a terrorist list, not everyone goes, wakes up guys, I’m going to be a terrorist. Okay. So, you know, sometimes people become terrorists, or they might be groomed into terrorism. So, if they joined a bank five years before they got groomed into terrorism, they would never appear on a terrorist watch list. So you’ve got to review your customers ongoing. So there has to be a way to do it in almost real time. So let’s change all the time sanctions change all the time. So governments change their sanctions list and who sanctioned sometimes on a daily basis. And also news changes. So if someone becomes a criminal or someone gets convicted of corruption or some becomes a political person, so someone gets appointed as a minister or within government, that can change on a daily basis. So you have to have a way to not checking onboarding throughout the entire lifecycle. So a nice way to do end to end management of that customer. And this allows you to do that.
Sure. So it’s, like, like you said, it’s a breakthrough in screening. And it’s a new solution, we launched a and the solution ultimately allows organizations to manage their screening risk, globally and across the whole organization. So what it allows organizations to do is to take in multiple data sources. So historically, and legacy screening solutions will often just look at one data source. So one list source, which looks at PEPS, which is politically exposed persons, and sanctions risk. So this is risks where countries, governments will put sanctions and restrictions on either countries or individuals or corporates, stating that these entities, individuals or countries cannot deal in US dollars or, or euros or other things. So you know, that this is a government-dictated list saying that banks cannot deal with these individuals, people. So legacy systems very much deal with historic lists, and it’s very structured in terms of the data they get, that creates inherent risk that certain data might be missed. And it doesn’t expand into the scope of things like adverse media, it doesn’t often expand into things like ships, some ships can be sanctioned, because they can come from sanctioned country, for example, like Syria, or Yemen. And so there’s a lot of risks that go way beyond just looking at politically exposed people, and individual and corporate sanctions. And we’ve broken down those barriers. So that risk by integrating multiple data sources, both in open source and premium data sources, into our streaming solution. So it gives the regulated organization, the power to screen across multiple data lists, both structured and unstructured, which gives them a lot more control over managing their risk as an organization.
Loraine Lawson
You’re using biometrics as well, in the solution and AI.Adam McLaughlin
Absolutely. So I think historically, with, again, you know, looking at Legacy screen solutions, and one thing I’d like to sort of highlight is transaction monitoring, and KYC procedures, you know, a part of the AML program, both systems have have already come into the world of AI. So they’re already utilizing AI and analytics to help enhance their detection, to help make sure that what they’re detecting is accurate and precise. But screening is often lag behind those screening solutions, Lexi have often not integrated this new technology to help enhance their screening, sort of accuracy. And so what we’ve done, we’ve been fused within the course of the platform, why we’ve infused AI, to help make sure that what’s been screened and the data that’s coming into the system, we’re able to accurately assess the right people, we’re able to make sure that the people we’re screening against, and the hits we’re getting are the same person. However, there is still a limit. So AI can do a lot and it will massively increase the accuracy. But sometimes you American text based screening, they’re still you know, there’s common names, like for example, across Europe, your Smith, for example, as a surname. So some common names, it’s still quite common in in various jurisdictions. And so you might get a very a lot of hits, coming back after disposition, those hits and say, which of these hits is the right person and actually, the person I’m screening is this a person related to these visa hits I’m getting from the list data. So biometrics takes it to that next level. So if you have a passport or a driving license or some photographic information of that customer that you’re screening against, You can feed that into the system. And it will it will check against a list of photographs. And these photographs are photographs of political people. So they might either be direct politicians, or they might be family members, politicians, they might be sanctioned individuals as individuals who have got a sanction against them, or they might be terrorists. So, you know, the solution has taken videos, news articles, and open source photographs of individuals into a database, so we can screen against that database. And if the photograph of the customer matches this database, it will return an accuracy score. And so what that can do is take the alphanumeric hit that you’ve got, and give very good precision, that the person you’re looking at is exactly the right person, because you’re actually going to photograph as well as alphanumericLoraine Lawson
text. Okay. And when it says financial services organizations use this, but when you see banks using it, what what would be their use case for this? I use a screening, but like in which situations?Adam McLaughlin
Sure, so screening is absolutely foundation of, of AML. And it’s the foundation of compliance. So banks and financial institutions have to do this in the regulated sector. So it’s not just banks, you know, insurance companies, and any organization that is regulated, have to do this activity. So it isn’t just the banks, you know, anything that deals in finance, and has to be regulated by fincen. And other regulatory bodies around the world, have to do screening. And so this will happen right at the beginning of a relationship with a customer. So before you even take them on as a customer, you have to make sure that they’re not sanctioned, because if they’re sanctioned, you cannot be dealing with that customer. So that is the end of the relationship.Loraine Lawson
To this network, we primarily write about banks. So how did they do it now.Adam McLaughlin
So at the moment that they do use wind solutions at the moment, but what they haven’t, what they don’t do is they don’t infuse AI. And very often, they just use singular lists. So they will use list providers. And there’s a number of them out there. But often banks will plug into one list provider to do their checking. And we’ve got one list and you don’t have the AI that sits behind it. What that generates is a lot of what we call false positives, that’s a lot of noise, if you like. So the sum of some figures we’ve gotten, there’s a report from eBay. And it says roughly that the number of false positives is about 99%. So the results, I get back, roughly 99% those results, I get back from systems without using AI to try and clean that data. You know, 99% of these hits are actually false, which creates a lot of work a lot of effort, and ultimately cost a lot of money for the banks to manage their screening programs, because they have to employ people to look at the results to reduce those 99 or workflows, 99% of false positives. So that’s how it works the moment and what we’ve done is we’ve taken change the curve, if you like and said actually, by utilizing AI, by utilizing biometrics, by ingesting additional data sources across that lifecycle, we can actually start enhancing that you accuracy and precision on making sure that the person you, your customer, your prospect is the right person who’s who matched against on a peps list or sanctions list or an adverse MEDIA list. And I just like to sort of go back to point so you do onboarding. So you have to make sure that from day one, you understand that the risk of that customer but it doesn’t stop there. You have to screen throughout the lifecycle because everyone changes. So for example, a terrorist list, not everyone goes, wakes up guys, I’m going to be a terrorist. Okay. So, you know, sometimes people become terrorists, or they might be groomed into terrorism. So, if they joined a bank five years before they got groomed into terrorism, they would never appear on a terrorist watch list. So you’ve got to review your customers ongoing. So there has to be a way to do it in almost real time. So let’s change all the time sanctions change all the time. So governments change their sanctions list and who sanctioned sometimes on a daily basis. And also news changes. So if someone becomes a criminal or someone gets convicted of corruption or some becomes a political person, so someone gets appointed as a minister or within government, that can change on a daily basis. So you have to have a way to not checking onboarding throughout the entire lifecycle. So a nice way to do end to end management of that customer. And this allows you to do that.
