When will you discover document analytics to uncover financial crime?

Discover how document analytics is helping to combat financial crime, including fraud and money laundering.

In brief

• Documents play a critical role in financial crime prevention such as Know Your Customer (KYC), anti-fraud and anti-money laundering.
• Document analytics can help financial institutions to quickly analyze documents and to detect manipulation.
• Large language models like ChatGPT are promising solutions to revolutionize document analytics.

From client onboarding to client montoring in the know-your-customer (KYC) lifecycle, loan applications to evidencing legitimate business interests: many processes in the financial world rely on supporting documents to confirm underlying information. Financial institutions need to consider a myriad of documents – electronic or physical – to run their compliance procedures, detect financial crime and respond to irregularities. Examples include:

• Bank statements and transaction records, providing records of financial activity and confirming net worth
• Invoices and purchase orders, which can be used to identify false billing schemes or other types of fraudulent activity
• Corporate records, documents such as articles of incorporation, bylaws and shareholder agreements, which can provide insight into the ownership and structure of a company
• Tax returns and financial statements, which can provide a comprehensive picture of a company’s financial health or an individual’s net worth
• Identification documents, like passports, used to confirm identities
• Purchase agreements and other commercial contracts governing the mercantile exchange
• Loading documents confirming cargo traversing the world via air, land or sea

KYC, anti-fraud, anti-money laundering and the fight against corruption are just some examples of the areas that depend on the analysis of documents as an information source. But the documents come in a wide range of formats and contain unstructured data which can be difficult to access and assess. The associated risks are twofold:

1. Risk of not being able to analyze documents as part of mass-monitoring processes
2. Risk that electronic documents may be manipulated and forged

In the absence of structured data, with its convenient tabular format, new data analytics approaches are needed to query, filter, aggregate and search the contents for potential wrongdoing. Technical solutions already exist and are continuously improving. In this article, we explore three use cases where data analytics is changing the value of documentation.

Use case: client onboarding

In the process of onboarding new clients, financial institutions are required to conduct due diligence checks to verify the identity and legitimacy of their clients. The process typically involves the collection and analysis of a wide range of documents, including passports, utility bills and financial statements.

Document analytics can play a critical role in this process by helping financial institutions to analyze the documents provided by clients quickly and accurately. For example, machine learning algorithms can be trained to identify key information such as names, addresses and account numbers, and to flag any discrepancies or potential issues with the documents. In addition, document analytics can also be used to confirm completeness of documents, e.g., confirming that all required forms are present in a client file, properly signed and consistent.

By leveraging document analytics in the KYC process, financial institutions can more effectively manage their risk exposure and comply with regulatory requirements. They can also streamline the onboarding process for their clients, reducing the time and effort required to gather and analyze the necessary documents.

Use case: anti-money laundering in trade finance

Trade finance involves the financing of international trade transactions. However, trade finance is also vulnerable to money laundering and other types of financial crime. At the same time, it is notoriously difficult to monitor due to its heavier dependence on documents than other types of transactions.

To combat this threat, financial institutions are increasingly turning to document analytics to identify potential money laundering risks in trade finance transactions.

Document analytics can be used to analyze bills of lading, which are documents that provide evidence of the shipment of goods. By comparing the details on the bill of lading with other documents such as purchase orders, invoices, and shipping manifests, financial institutions can identify any discrepancies that may indicate the use of false or fraudulent documents. They can also identify any unusual patterns or behaviors, such as shipments to high-risk countries or the use of complex corporate structures to conceal the true ownership of the goods being shipped.

In addition, document analytics can also be used to analyze financial documents such as letters of credit, which are used to facilitate payment for trade transactions. By analyzing the details of the letter of credit, including the terms and conditions and the parties involved, financial institutions can identify any potential risks or red flags that may indicate the use of the letter of credit for illicit purposes.

Use case: detecting manipulation of electronic documents

Document analytics is an essential tool in anti-fraud efforts. Fraudsters often create fake documents to support their schemes, such as falsifying invoices or receipts to make it appear that expenses were incurred when they were not. In other cases, fraudsters may alter legitimate documents, such as changing the amount on a check or altering a contract to benefit themselves.

Detecting electronically manipulated documents typically involves using specialized software tools that are designed to identify signs of tampering or manipulation. There are several techniques that can be used to detect electronically manipulated documents, including the following:

Metadata analysis

Many electronic documents contain metadata – information about the document that is not visible to the user. Metadata can include information about the software used to create the document, the date and time the document was created, and other details. By analyzing the metadata associated with an electronic document, it is possible to identify signs of tampering or manipulation.

Digital forensics

Digital forensics involves analyzing the digital artifacts left behind by electronic documents. This can include examining the file structure of the document, looking for signs of alterations or modifications. Digital forensics can also involve analyzing the device used to create or modify the document, e.g., looking for signs of malware or other unauthorized software.

Image analysis

If a document includes images or photographs, image analysis tools can be used to detect signs of manipulation. This can include analyzing the pixel patterns in the image and looking for inconsistencies or artifacts that may indicate that the image has been altered.

Machine learning

Machine learning algorithms can be trained to detect patterns in electronic documents that may indicate manipulation or tampering. For example, a machine learning algorithm may be trained to identify common techniques used to alter electronic documents, such as copying and pasting or changing the font size.

Financial institutions may use these techniques in isolation or in combination to verify the authenticity and integrity of documents submitted by clients or to identify potential cases of financial crime such as insurance claims fraud, loan application fraud or submitting falsified source of funds documentation.

Outlook

Although documents are increasingly being digitalized and easier to analyze electronically, paper versions are still here to stay. The legacy of recent decades alone will mean that we will continue for some time to rely on information extracted from documents and will need to confirm their authenticity.

The advent of large language models like ChatGPT has the potential to revolutionize document analytics in financial crime prevention. Over the last six months, we analyzed the potential with mockup data during initial tests. It provides some insights abouts the future potential. However, the technology also comes with significant privacy concerns. Financial institutions will have to either accept uploading their most sensitive data to external service providers or invest in a – costly – private cloud solution. On the flip side, large language models also make it easier than ever to create fake but plausible sounding documents, from invoices to targeted phishing emails.

Whatever developments emerge, the use of document analytics, document authenticity tools, and other specialized software tools is set to grow as financial institutions realize their potential in analyzing documents more efficiently and accurately. By leveraging these tools, financial institutions can effectively manage their risk exposure, comply with regulatory requirements and streamline their processes.