Deepfake AI and the Future of Document Verification

The increasing accessibility of artificial intelligence (AI) technology poses a new threat: individuals with malicious intent can now more easily create realistic deepfakes that are challenging to differentiate from authentic content, amplifying the threat to businesses and individuals alike. Traditional methods of document verification often merely conduct a visual inspection of a document, leaving organizations vulnerable to fraud losses and reputational damage.

The pandemic and subsequent shift to digital-first customer acquisition and account management has caused a significant uptick in demand for effective document verification technology within the past 12-18 months. According to The Forrester Wave™: Identity Verification Solutions, “The increased focus on optimizing customer experience and reducing digital friction is causing organizations to migrate away from knowledge-based IDV [(identity verification)] (credit file headers) to IDV methods that rely on physical-ID-document verification and are phone-number-reputation-based.”

Organizations looking to streamline their omnichannel customer experience with digital document verification need to be sure they’re not creating more manual reviews due to false positives. This requires robust identity verification capabilities that include a multi-layered approach to document verification.

What is document verification?

Document verification is the process of verifying the authenticity of a document to determine that it is genuine and has not been altered in any way. The document verification process can involve several steps, including checking the document’s details against a database of known fraudulent documents and examining the document’s physical characteristics to confirm it matches the standard template from the issuing authority.

Document verification just one step in identity verification

Document verification, by itself, is not identity verification or identity proofing, but rather only one important step in the identity proofing process. What is the difference?

• Document verification is a digital identity verification method used to check whether an applicant-provided photo ID (e.g., passport, ID card, driver's license) is legitimate. The goal of document verification is to capture, extract, and analyze identification data to authenticate government-issued identity documents.
• Identity verification or identity proofing, on the other hand, is confirming or denying that a person is who they claim to be by comparing the credentials (something you know, something you have, something you are) of a person requesting access with those credentials previously proven, stored and associated with the identity being claimed. This process not only includes document verification, but often also includes biometric verification as well as behavioral verification. In other words, document verification alone is not identity verification.

Fake and tampered-with documents can outsmart document verification technology

Fraudsters are adept at passing off phony documentation to open accounts and complete credit applications, even to the point they can fool document verification technology. Documents, like Internal Revenue Service (IRS) forms, identification cards, business incorporation documents, can be faked using AI technologies. Backup document verification, like driver’s license verification can be fooled using fake selfies and driver’s licenses. Increasingly, criminals execute organized fraud attacks at scale, where they re-use the same information, submitting hundreds of documents with the same face or similar document numbers.

Given fraudsters’ ability to create fake documents to fool verification technology, organizations need to think more holistically about identity proofing and document verification technology. Repeat fraudsters can be caught by failing document and biometrics checks, but also by an array of passive digital signals. This means that one can identify many fake documents by using IP address or device identifiers. 

What to look for in a document verification solution?

While evaluating document verification solutions, be sure to look for capabilities that do not rely exclusively on visual inspections of the document. Combatting AI requires a solution that leverages multiple signals and checks that enable it to uniquely stand up against deepfake technology:

• Run fraud checks against PII: Confirm that the personally identifiable information (PII) data provided matches against consumer credit and non-credit identity databases.
• Confirm with device reputation checks: Triangulating document verification with checks against a device reputation consortium to understand if the device being used has been associated with fraud in the past
• Compare document photo to real-time selfie: Request a selfie from the user to validate a match, supported by liveness detection technology that prevents system manipulation.
• Combine document and biometrics: Use (selfie + liveness) to set higher thresholds for validation. Motion is significantly more effective than still images at preventing fraud.
• Proactively improve algorithms to detect deepfakes: Continuously researching and implementing preventative measures to block deepfake-based fraud, including images that are presented from digital screens, as well as algorithms that can detect when a document is printed on paper, instead of legitimate document material.

What’s next?

Assessing the risks associated with an identity starts with a better understanding of what technology is required to meet unique identity verification needs, and by implementing identity verification solutions robust enough to keep fraudsters at bay while providing customers with the quality experience they deserve.