Facing the challenge of fraud and money laundering head-on

"Technological innovation constitutes a tremendous growth engine for building a better world for us all. Despite the message of progress, humans have remained human and unfortunately, criminal elements choose to take advantage of these advanced capabilities and work methods for fraud and money laundering,” says Dr. Ami Appelbaum, Innovation Authority Chairman and Chief Scientist in the Ministry of Economy & Industry. ”The Innovation Authority is glad to be a partner in the effort to contain this activity, among others by supporting outstanding startup companies operating in the fintech field and assisting us all to preserve a safe financial environment".

"Preventing financial crime is a multi-billion-dollar industry", says Yuval Marco, General Manager of the Fraud and Authentication Management Business Unit at NICE Actimize. According to Marco, in 2020 alone, banks and financial organizations were fined more than 10 billion dollars by various regulators for insufficient effort and underachievement in preventing money laundering. The losses as the result of fraud are even greater: 50 billion dollars just from identity fraud, i.e., when criminals steal the identity of others and take over their bank accounts or create fictitious identities and open accounts via which they conduct their criminal activity.

"These numbers are very large, and, for several reasons, we believe that this criminal industry will only continue to grow", says Yuval Marco. "The first reason is that the crime organizations are becoming increasingly sophisticated – and they also use Artificial Intelligence and machine learning".

The second reason Marco mentions is connected to the leading trend of recent years and one which has grown during the Covid crisis – the accelerated transition to the digital world. Financial services organizations have adopted a strategy of moving rapidly to digital, and a process that was supposed to take several years has happened very quickly. Financial organizations now allow more customers more digital possibilities. One can even apply digitally to open an account with a new bank without being physically present even once. "On one hand, this offers benefits for the customer but on the other hand, it also creates risks for the financial institution", Marco says. "We can already see the beginnings of financial crime changes, and this is expected to continue in the future.

Another significant trend pointed out by Marco is that of faster payments. Today, money can be transferred to another person at the click of a button, and the funds become available for immediate use in the recipient's account. This means that the time the banks have to analyze the transactions and to ascertain whether it poses a risk has been significantly cut down to milliseconds. This necessitates a transition to advanced technology that can analyze the risk in a fraction of a second.

According to Marco, all these processes lead to better customer experience, more business opportunities for the financial institutions, but also more crime and more money that needs protecting. This situation is exacerbated by the fact that unlike financial institutions that adhere to regulation, crime organizations make use of sophisticated technologies without any regulatory or ethical inhibitions.

A Market of Fraud Prevention Solutions

NICE Actimize supplies financial crime and compliance solutions to organizations of all sizes including the world’s largest global financial organizations, for preventing money laundering, fraud, and market manipulation. The company – founded more than 20 years ago – was subsequently acquired by another Israeli company – NICE. "If you look at NICE as the parent company, Actimize is the division that focuses on outsmarting financial crime with intelligent solutions", Marco explains. NICE Actimize has more than a thousand employees and we serve customers located in more than 70 countries around the world.

"We presently lead the market and work with approximately 750 financial organizations worldwide, including the leading ten global financial services organizations, and we cater to the full range of needs, from packaged, fast to deploy SaaS solutions to customized solutions including self-development capabilities for the more mature organizations", says Marco.

"Our solutions are installed at the financial organizations themselves or in the cloud and our platforms continuously receive information about every transaction that the institution executes", Marco explains. "We also receive ongoing relevant data to identify financial crime such as lists of criminal elements or other supplementary solutions that provide additional information, for example, about the device from which the transaction was executed – who it belongs to and whether certain apps or software are installed on it.

"We run the analytical models and AI with the aim of informing the organization of the risk level for each transaction. "If we identify a risk of money laundering or fraud, we can disrupt the transaction, step-up the authentication (e.g., send one-time-password to a trusted device) and alert the company of the suspicious activity", Marco clarifies. "We also provide them with intelligent & automated tools in order to complete the investigation of suspicious accounts. The outcome of the investigation (ascertaining whether it is fraud or not, which type etc.) is being used to continuously optimize our models and improve the precision.

"In the world in which we live, the more data you have, the better job you can do", Marco explains. "But to obtain the data from different sources, transform, correlate, and integrate it, takes a lot of energy and time. In practice, studies in this field show that about 40 percent of the organization's fraud prevention expenses are spent on the acquisition and integration of data sources, and on making them operational.

Understanding of this problem, the company launched X-Sight Marketplace, a cloud ecosystem of sellers and end customers – just like the Google Store. Customers can choose the solution they want to connect to their NICE Actimize solutions.

"Part of the innovation here is the advantage in using the cloud to save our clients the hassle of integration. It significantly enhances the ability of these organizations to add information and intelligence to our platform in order to improve their identification", says Marco. "The Marketplace is part of a broader platform called X-Sight. Every 1-2 weeks, we release new features that enable us to cover a whole world of complementary solutions for preventing money laundering and fraud. For example, when we saw criminals creating accounts and using them for a whole range of purposes, we launched a specific solution".

One fraud method that has developed rapidly in the US over the past year is that of synthetic identity fraud – in other words, criminals steal different identities and use them to create a mixture – one person's name, another person's address etc. To identify such cases, NICE Actimize checks when and where else a certain identity was seen and its level of credibility. Was the user's email address created only recently? Does the same identity appear on the Dark Web – the hidden internet network that is used by dubious parties and where identities can be bought? We do this by collaborating with companies that specialize in these specific areas".

​ 

How to Bypass the Problem of Privacy?

"Questions are raised in the world of analytics about how to use Artificial Intelligence to identify risk. This is the point to mention two concepts: Federated Learning and Online Incremental Learning", Marco says.

When criminals see that a scam or fraud works, they replicate it at other banks so there is tremendous value in the sharing of indications between our clients as part of AI. "Our clients' data flows to the cloud but due to regulation and their cybersecurity requirements, each of their customer's data is stored separately. The Federated Learning method enables us to take a model from Customer A and use it as part of a model that we provide to Customer B", Marco explains. "In this way, we can share data without exposing the customers".

Danny Butvinik, the company's Chief Data Scientist, adds: "We can't ascertain the identity of the specific customer from the insights we gain. These are merely parameters, not names, but these variables act as the foundation for building models that serve everyone. In this way, everyone gives a little but receives a lot so that every client of NICE Actimize, whatever the size of their company, can benefit from this method".

NICE Actimize's clients are financial institutions: financial managers, traditional or digital banks, and the fintech industry which provides financial services to consumers and corporations that are the target of criminals. These organizations have entire divisions that tackle financial crime. They have teams responsible for fraud and other teams in charge of working on money laundering. One of the trends that NICE Actimize is advancing is synergy between these two groups because of the direct connection between them, something that the large organizations are also now beginning to understand".

"With fraud, the damage to both the bank and the customer is obvious. In the world of money laundering, financial institutions must comply with regulations, and they want to best serve their customers and prevent societal issues like human trafficking, arms trade, drug trafficking, and more because these are the types of organized crime rings that money is laundered to. If a bank does not adequately comply nor stop money laundering the down-stream impact is massive fines and reputational damage.

Butvinik tells of their very close collaboration with the Innovation Authority: "The goal is to create innovation that differentiates us from other companies. In addition, this innovation must connect directly to our analytical needs at the company – innovation that can be seen in our products and with our clients. These are two very important aspects that contribute to our successful collaboration with the Innovation Authority. 

"For example, this year we are working on an innovative project of online digital learning and are aspiring to achieve maximum possible automation and

speed. We want to know in real-time about trends, risks, and other information without any indication that we are doing so, as is the case with most companies worldwide".

"Our strategy is very clear", Marco summarizes. "And it has a name: 'Autonomous Customer Lifecycle Risk Management'. The goal is to change the paradigm from a reality whereby the fight against economic crime is siloed, labor intensive and reactive – with multiple teams that need to maintain and operate these programs, to a world in which we are leveraging advancements in AI, cloud technology, and automation to use machines for the "heavy lifting" of fighting financial crime and have humans merely "correct course" when necessary.

We intend to implement this across the customer lifecycle – from the decision whether to accept a new customer, into the ongoing monitoring of accounts, through to offboarding when a customer has the right to erase all their account information. We believe that providing a holistic view of the customer and the associated risk is integral to stopping financial crimes.

Looking for the Unknown Unknowns

In 2013, Prof. Amir Averbuch founded ThetaRay, a fintech company together with Prof. Ronald Coifman. Today, he serves as the company's CTO. ThetaRay's goal is to prevent financial crime by locating big data anomalies in banks and financial institutions that may indicate crimes such as financial fraud, money laundering, funding of terror, and drug and human trafficking.

"The anomalies we identify can originate in three verticals: IOT (The Internet of Things), cyber, or financial", Prof. Averbuch explains. "We are presently focusing on the financial field and our target market is financial institutions where money transfers are executed, such as banks, fintech, and blockchain companies that offer payment transfers. Many large banks around the world are already our clients. We can offer solutions to banks and large financial institutions as well as to the SWIFT industry, such as small banks and private organizations that offer international money transfers, for example to migrant workers who send money home. When you transfer money to someone, each bank has its own SWIFT code which is included in the financial transaction".

ThetaRay is built on three pillars: algorithms, a platform, and data sciences. These three pillars enable the company to offer complete and coherent solutions for all types of financial institutions obligated by regulation to adopt a solution that will identify semi – and un– supervised financial crime in data.

The algorithms developed by the company are the core of the system and they are suited to the three verticals. These algorithms use geometric methods and extremely advanced deep learning and Artificial Intelligence technologies. "We were among the first companies to build a system for identifying financial crime that runs on AI infrastructure", says Prof. Averbuch. "Our system is the only one in the world that simulates human intuition or gut feeling. We call it 'Artificial Intuition'.

"Our big advantage is in the development of algorithms that don't use rules. In practice, we don't know in advance what we're looking for. Whereas algorithms that conform to rules need to know what to look for in order to write a rule, we are searching for the unknown unknowns. One disadvantage of working according to rules is the creation of a large number of false alarms".

ThetaRay's solutions come in three primary forms: on the client's premises ('On Prem'), in a cloud, and 'Software as a Service' (SaaS). They detect 95% of worthy anomalies compared to other solutions that offer only 5%-30% detection. The number of false alerts of ThetaRay's solution is 95% lower than the old tools and, accordingly, they reduce the time spent investigating alerts by at least 50%. The average time required for assimilating such a solution stands at only 6 weeks as opposed to 6-12 months.

To better understand how ThetaRay's system works, Prof. Averbuch adds that: "We developed a large number of algorithms. Each algorithm undergoes a stage of learning the data. These construct a model of normality with each algorithm building a different kind of normality. We use multiple algorithms to increase the number of genuine alerts and to reduce the number of false alerts. The final reported anomaly is not determined by a single algorithm but rather, by the fusion of all the participating algorithms".

ThetaRay's innovation is the platform that implements the algorithms. "We have two platforms – a large-scale one that is sold to the big banks and a smaller, more compact platform that operates in the cloud and is suited to the cross-border SWIFT payments market", explains Prof. Averbuch. "We have built a platform with the most modern software components that can either connect to the cloud or not, that runs our algorithms, and which finds all the anomalies that point to financial crime.”

The method developed by ThetaRay enables financial institutions to find unknowns that you didn't even knew existed, with very few false alerts – 95% less than other existing systems. The investigative reconstruction gives the user a lot of forensic information that helps in finding the problem quickly. The platform is a unifying one, it can communicate with the cloud and has the option of a graphic user interface, both for entering data and for the results so that you can create a graphic geometric interaction with the data.

The third pillar is the company's data scientists. "Our data scientists help the client apply the system, match the features that each client needs, and prepare the data for them", Prof. Averbuch explains. "With large banks, this pillar involves a lot of work but less with the SWIFT systems. In the future, we hope to completely remove the data personnel's intervention in the SWIFT field and make it entirely automatic".

When observed as a process, ThetaRay's algorithms find the suspicious transactions, fuse them, and point out the anomalous data points. This is all determined by the platform. "We are not satisfied with just identifying the anomaly but want to find a large amount of data that enables the client to perform the forensic analysis and understand what occurred with this anomaly", says Prof. Averbuch, who adds that "naturally all this occurs in real-time".

"In many places, we found anomalies that signaled the start of money laundering, even before it actually happened and indeed, several months later it turned out that this was the case. We challenge those places where the criminal infrastructure is just getting organized.

"ThetaRay's innovation is expressed in the very special algorithms it has developed, many of which are protected by patents. The approach that we have developed for identifying anomalies is unique, the most innovative and advanced there is, and utilizes the range of very complex tools that we developed ourselves and our professional teams of data engineers”.

In 2013, the company was a small startup with ten employees. Today it has about 80 employees and maintains offices in Israel, the US and UK. Our client base is also global.

Hand in Hand with Regulation

To succeed in the complex mission of protecting financial institutions from crime, we don't necessarily need to invest more resources but rather, to construct an integral policy of government entities and smart regulation. ThetaRay operates hand in hand with regulators globally, as well as with the banks to develop AI-based regulation that is suited to the information explosion age.

According to Prof. Averbuch "ThetaRay has introduced new advanced technology that allows the company genuine control and prevention of the spread of financial crime”. Prof. Averbuch adds that "ThetaRay’s forensic technology is not a 'black box'. It enables banks and regulators to trace the root cause of an anomaly and understand what happened in the transaction".

"When examining rule-based systems, you discover departments with thousands of employees who are required to check every alert and the suitability to each law. These departments are extremely large because the old regulation demands that everything be checked. The result is very high overheads for the banks. ThetaRay, in contrast, reduces the banks' overheads and enhances the quality of control and the adherence to regulatory demands in the shortest possible time", says Prof. Averbuch.

What Does the Future Look Like?

"If you take weather forecasting as an example, today's meteorologists are only able to predict the weather 3-7 days in advance", says Prof. Averbuch. “The tools that currently exist combine the updated real-time data and scientific models of equations for change in the weather, enabling them to accurately forecast the weather several days ahead. The next stage will be the transition to building a genuine future forecast that is not based on the past but rather, on the internal understanding of what leads the weather. It needs to be invented but it's the future”.

According to Prof. Averbuch, the next stage will be to define with 100% accuracy what is unusual and what is not, in order to achieve reliable diagnostics. Autonomous driving systems such as Mobileye can be taken as an example. The next stage will be when our cars are "frightened" by a piece of paper that flies onto the road, leading them to slow down and cause an accident. Meanwhile, they will update all the other cars on the road and teach them from experience, thereby transforming them into a learning community or group of computers that talk to each other about situations they are stuck in.

ThetaRay has already embarked on the next stage by developing a unique technology – Intuitive AI. This technology can define the dangerous situations and warn that something is not in order, like a person hearing a sound from the car engine and intuitively understanding that something is wrong.

"In the field of medicine today, you can receive a second opinion from computers but not the initial diagnosis. Years more of development are required to achieve an accurate initial computerized diagnosis, in my opinion", Prof. Averbuch claims. "In the future, we will see more AI-based systems helping companies to protect themselves. We will even see solutions that until recently were considered impossible, such as intuitive AI that mimics the human decision-making process. Such systems are already in use in global banking systems as part of our program to fight money laundering or to expose fraud. The market will realize that the only way to protect itself is to use the most advanced solutions possible".

 

The article was written in collaboration with the Israel Innovation Authority, responsible for the country’s innovation policy. Its role is to nurture and develop Israeli innovation resources, while creating and strengthening the infrastructure and framework needed to support the entire knowledge industry.