Why are financial crimes on the rise? Many people ask this question as crime-cases in the financial industry rise. Banks according to a McKinsey report¹ have lost millions of dollars in the last decade alone and this could worsen as criminals upgrade their financial crime tactics. Financial crime analytics can help financial institutions, investigators detect fraud, and money laundering, assess risk, and report on data to prevent financial crime.

Each year, many cases of banking fraud² increase and despite stringent measures, losses continue to spike with financial institutions lacking concrete strategies to address this growing problem. Analytics help to pinpoint transactions that need further scrutiny, identifying the needle in the haystack of financial data.

With only a 1% success rate in recovering stolen funds, the financial services industry has realized that traditional approaches to dealing with financial crime are not working. Across the ecosystem, regulatory authorities, enforcement agencies, and financial institutions³ are working together to disrupt financial crime. This requires a proactive approach to predict and manage the risks posed to people and organizations and not merely to comply with rules and regulations.

The challenges faced by financial institutions regarding money-laundering activities have increased substantially in the globalization era. Additionally, there is a rising menace of financial crime and counterfeiting. As money launderers become more sophisticated, the effectiveness of anti-money laundering policies is under heightened regulatory scrutiny. The probability of banks facing rigid penalties and reputation loss in case of shortcomings in AML management has increased.

A good example of a tool used for financial crime detection is AMLOCK. This is the enterprise level end-to-end financial crime management solution. It integrates the best of anti-money laundering⁴ and anti-fraud measures to effectively identify, manage, and report financial crime. It provides various features that cater to profiling, risk categorization, transaction monitoring, and reporting requirements of financial institutions. Features that form part of this offering are in line with AML (Anti Money Laundering) regulations.

In this article, I will explore current practices in financial crime detection, use cases and explore what the future looks in financial technology and fraud reduction.


Criminals are pervasive in their determination to identify and exploit vulnerabilities throughout the financial services industry. Their ability to collaborate and innovate necessitates a proactive approach towards responding to individual events, while disrupting crime networks. Combating #financialcrime is complementary to generating revenue. The big data analytical capabilities that enable a bank to personalize product offerings also underpin an effective approach to spotting and responding to criminal behavior.

To out-pace fraudsters, financial institutions and payment processors need a quicker and more agile approach to payment fraud detection. Instead of relying on predefined models, applications need the ability to quickly adapt to emerging fraud activities and implement rules to stop those fraud types. Not only should organizations be able to adjust their detection models, the models themselves should be inter-operable with any #datascience, machine learning, open source and AI technique using any vendor.

In addition, to eliminate fraud traveling from one area or channel to another undetected, aggregating transactional and non-transactional behavior from across various channels providers greater context and spots seemingly innocuous patterns that connect complex fraud schemes.

Artificial Intelligence For Financial Crime Detection

Within financial institutions, it is not uncommon to have high false-positive rates that is, notifications of potential suspicious activity that do not result in the filing of a suspicious transaction report. For AML alerts, high false positives are the norm.

The reason for this is a combination of dated technology and incomplete and inaccurate data. Traditional detection systems provide inaccurate results due to outdated rules or peer groups creating static segmentations of customer types based on limited demographic details.

In addition, account data within the institution can be fragmented, incomplete and housed in multiple locations. These factors are part of the reason why alerts and AML are key areas to apply #artificialintelligence, advanced analytics and RPA.

The technologies can gather greater insight, understand transactional patterns across a larger scale and eliminate tedious aspects of the investigation that are time-consuming and low value. AI can augment the investigation process and provide the analyst with the most likely results, driving faster and more informed decisions with less effort.

AI based Intelligent Customer Insights

Periodic reviews of customer accounts are performed as part of a financial service organization’s risk management process, to ensure the institution is not unwittingly being used for illegal activities. As a practice, accounts and individuals that represent a higher risk undergo these reviews more often than lower-risk entities. For these higher-risk accounts, additional scrutiny is performed in the form of enhanced due diligence.

This process involves not only looking at government and public watch list and sanctions lists, but also news outlets and business registers to uncover any underlying risks. As one would think, such less-common investigations took the majority of the due diligence process because they typically required lengthy, manual searches and validation that a name was the individual or entity under review.

With modern technologies like entity link analysis to identify connections between entities based on shared criteria, as well as #naturallanguageprocessing to gain context from structured and unstructured text, much of this investigation process can be automated. By using AI to perform the initial search and review of a large number of articles and information sources, financial institutions gain greater consistency and the ability to record the research results and methodology.

Much like the AML alert triage example previously mentioned, the key is not to automate analysts from the process. Instead, AI automates the data gathering and initial review to focus the analysts on reviewing the most pertinent information, providing their feedback on the accuracy of those sources and making the ultimate decision on the customer’s risk level.

Analytics for Financial Fraud Detection

Innovation in the payments space is at a level not seen in decades. From mobile payments, to peer-to-peer payments⁷ to real-time payments, there are a growing number of payment services, channels and rails for consumers and businesses alike. But these myriad options also give fraudsters plenty of openings for exploitation, as well.

Easy-to-exploit issues with these new payment services include their speed and lack of transactional and customer behavioral history. These issues put financial institutions and payment processors in a difficult position. If they block a transaction, they could negatively impact a legitimate user, leading the user to either abandon the platform or use a competitor instead.

If the transaction is approved and it is fraudulent, it erodes trust in the payment provider and leads to a loss. Traditional fraud detection systems were designed for a relatively slow-moving fraud environment. Once a new fraud pattern was discovered, a detection rule or model would be created over a matter of weeks or months, tested and then put into production to uncover fraud that fit those known fraud typologies.

Obviously, the weakness of this approach is that is takes too long and relies on identifying the fraud pattern first. In the time it takes to identify the fraud pattern, develop the model and put it into use, consumers and the institution could experience considerable fraud losses. In addition, fraudsters, aware of this deficiency, can quickly and continuously change the fraud scheme to evade detection.

Case Studies of Financial Crime Technology

Let us now explore some use cases of financial technology and how companies benefited in fraud reduction.

1. MasterCard

To help acquirers better evaluate merchants, MasterCard created an anti-fraud solution using proprietary MasterCard data on a platform called MATCH that maintains data on hundreds of millions of fraudulent businesses and handles nearly one million inquiries each month. As the volume of data in its platform grew over the years, MasterCard staff found that its homegrown relational database management system lookup solution was no longer the best option to satisfy the growing and increasingly complex needs of MATCH users.

Realizing that there was an opportunity to deliver substantially better value to its customers, MasterCard turned to the Cloudera Enterprise Data Hub. After successfully building, integrating, and incorporating security into its EDH, MasterCard added Cloudera Search and other tools and workloads to access, search, and secure more data.

2. United Overseas Bank (Asia)

The challenge UOB faced was the data limitations of their legacy systems. With legacy databases, banks are restricted by the amount of data as well as the variety. As a result, they miss key data attributes that are necessary for anti-money laundering, transaction monitoring, and customer analytics engines to work effectively. UOB established the Enterprise Data Hub⁸ as the principal data platform that, every day, ingests two petabytes of transaction, customer, trade, deposit, and loan data and a range of unstructured data, including voice and text.

3. Bank Danamon (Indonesia)

Bank Danamon is one of Indonesia’s largest financial institutions, offering corporate and small business banking, consumer banking, treasury and capital markets. Bank Danamon uses a machine-learning platform for real-time customer marketing, fraud detection, and anti-money laundering activities. The platform integrates data from about 50 different systems and drives machine-learning applications. Using #machinelearning on aggregated behavior and transaction data in real time has helped Bank Danamon reduce marketing costs, identify new patterns of fraud, and deepen customer relationships.

This is the Best Time to Implement AI for Financial Crime Detection

Financial crime and corruption are at epidemic levels and many countries are unable to significantly reduce corruption. Regulators and financial institutions are looking to innovative AI technology to fix problems that have grown beyond their ability to solve with intuition and existing tools alone. To justify cognitive initiatives, financial services organizations need to show real return on value in such investments.

IBM is able to demonstrate the value in a variety of use cases, as shown in the client success stories outlined in this white paper. A misunderstanding about artificial intelligence is the belief that it will replace employees. However, the financial crime analyst is and should always be an essential part of this process. AI, process automation and #advancedanalytics are tools that can perform analyses and tasks in a fraction of the time it would take an employee.

Yet, the ultimate decision-making power still lies with those analysts, investigators and compliance officers for whom this technology provides greater insight and eliminates tedious task work. This augmented intelligence is the next phase of the fight against financial crime, and one that only together financial institutions, regulators and technology partners can win.

Works Cited

¹McKinsey Report, ²Banking Fraud, ³Financial Institutions, ⁴Anti-Money Laundering, ⁵Payment Fraud Detection, ⁶Advanced Analytics, ⁷Peer-to-Peer Payments, ⁸Enterprise Data Hub