Document fraud has been a problem for as long as there have been written contracts. Legislation prohibiting forgery dates back to 80 B.C., when Roman law outlawed the falsification of documents to transfer land to heirs. Throughout history, fake or altered documents have been used to defraud unsuspecting buyers and the public, to promote a political agenda, to falsify identities, or to create misinformation. The better the forgeries, the more difficult they are to identify as phony.
Losses from falsified or forged documents run high. According to the U.S. Department of Justice, settlements under the False Claims Act exceeded $2 billion in 2022. Fifteen percent of all insurance premiums are lost to fraud, amounting to as much as $30 billion annually. Bank losses from falsified accounts and identity theft increased 109% in 2021.
As many as 50% of new bank accounts opened in 2021 were fraudulent, resulting in losses of $163 billion from stolen unemployment benefits and Paycheck Protection Program (PPP) loans. Additionally, according to the CoreLogic National Mortgage Application Fraud Risk Index, fraudulent home loans in the fourth quarter of 2022 were up 3.6% over the third quarter.
In a world rapidly adopting AI to automate tasks such as document processing, it is not surprising that many important documents no longer receive human inspection. To combat document fraud, automation technologies like intelligent document processing (IDP) may apply artificial intelligence (AI) and machine learning (ML) to authenticate documents. Verifying documents is now exceedingly important, and more organizations are using IDP powered by AI/ML to not only automate document processing but also use AI/ML validation to reduce the potential for fraud.
Identity theft is a significant part of the problem. Cybercriminals often target bank statements, Social Security checks, W-2 forms, and similar documents to generate false identities. Some of the most common types of document fraud include:
- Image fraud – A fake image can be used to substitute an image in a legitimate form of ID, such as a driver’s license or passport.
- Modified documents – By taking genuine documents and manipulating them by changing the names, dates, and ID numbers, fraudsters can use those modified documents to open a line of credit, take out a loan, or do some other crime.
- False documents – A more sophisticated form of identity theft, criminals can falsify documents, passing off someone else’s credentials as their own by changing account numbers, birth dates, and other identifiers.
- Illegitimate documents – Unlike false documents, illegitimate documents are fake and created to look like bona fide original documents, such as fraudulent checks, money orders, or invoices.
Automating Document Processing
It’s not always easy to identify fraudulent documents. Manual inspection may reveal falsifications, such as the lack of a watermark or other telltale clues to show a document isn’t genuine. However, manual inspection isn’t always accurate or practical, especially when processing thousands of documents at a time. Automating document validation is a more precise and efficient approach.
IDP simplifies document fraud detection through automated document analysis and digitization, converting paper documents to digital form using AI/ML to analyze, contextualize, and categorize document data.
IDP should not be confused with optical character recognition (OCR), which converts text into a machine-readable format. IDP is more robust than OCR, as it also indexes and categorizes scanned images and even analyzes documents for forgeries or fraud.
Rather than manually processing each document, IDP extracts information and categorizes it. It should also authenticate data, ensuring the documents are genuine and haven’t been altered before IDP intake.
While IDP automation simplifies processes and reduces the need for manual inspection, additional protocols are still needed to authenticate documents as part of digital intake. Without an accurate, authenticated original, fraudsters can easily alter digital documents.
Using AI, computer vision, and other technologies, documents can be analyzed for authenticity and fingerprinted. For example, AI can reveal possible alterations in text, and embedded images and photos can be examined for pixel manipulation and potential changes, such as forged signatures.
To prevent tampering, each digital document can be given a unique identifier or fingerprint to verify authenticity. Each fingerprint can then be anchored to a blockchain to avert insider or outsider document alterations.
Using IDP to process sensitive documents is proving to be a boon to financial institutions, insurance companies, healthcare providers, loan underwriters, and other organizations looking to save time and money. However, automation can risk increasing the amount of fraud if it is not coupled with automated fraud protection technology. Organizations worried about fraud and identity theft should use care in picking a solution that uses AI to speed and automate the processing of documents, coupled with AI fraud detection to replace the human inspection that was an integral part of manual processing.
Including fraud detection as part of IDP enables organizations to flag suspicious documents in real-time for further review only when potential fraud is detected, thereby not sacrificing automation efficiency. The result is tamper-proof document processing that saves time, money, and potential losses while safeguarding sensitive paperwork.