Photo ID Anti-Tampering

Learn about new IDScan digital anti-spoofing algorithms that are used in the Photo ID Match process.

Overview

The FaceTec Device SDKs scan Photo IDs and have AI checks to determine the Photo ID Document is a physical document and not a digital document on a screen.

Important Note: Digital IDs are already being used in some places, such as Colorado, USA, so it is not necessarily a spoof if an ID is shown on a screen.

As part of the Onboarding process, high-resolution digital images of the User’s ID Document are taken. These photos are processed on the User’s device; a proprietary, tamper-proof payload is created and encrypted by the Device SDK. Your App sends the IDScan payload up to your Server, where it is decrypted by the FaceTec Server. The FaceTec Server verifies the images contain the entire ID Document, and if they don’t, a retry request is sent to the User, and if the IDScan was captured from a digital computer screen, rather than being a physical document, it will be flagged as "Likely Spoof."

State-of-the-Art Onboarding

FaceTec changes the order of operations of the Onboarding process to increase usability, security, and reduce costs. By performing the Liveness Check first, the Users move further through the sign-up process before they have to go retrieve a document, and since fraudsters don’t want to put their 3D Face on camera, they never even get the opportunity to show Fake IDs to the system.

Requirements

Your App must utilize FaceTec's Photo ID Scan User Interface, which is included in all of the FaceTec Device SDKs and has features to perform some basic anti-photo swap and anti-tampering checks.

Overview of Anti-Spoofing & Anti-Tampering Checks

There are several new behaviors, properties, and checks that FaceTec performs during the ID Scan and Photo ID Match:

  • The Photo ID is checked to ensure it is not a digital spoof. A digital spoof of an ID is an ID presented on a digital screen.

  • The ID Scan is checked to ensure a Full ID is presented. The ID may be deemed "full" if the vast majority of the ID is present. If 4 corners of the ID are in view, the ID should be deemed "full".

  • A Match Level will not be returned if FaceTec assesses the ID Scan was not a complete ID.

  • The Photo ID is checked to ensure the face region has not been replaced or tampered with. This will detect the replacement of the photo ID image in either a digital artifact or a physical ID.

  • The Photo ID is checked to ensure the text on the document has not been replaced or tampered with. This will detect the replacement of the text in either a digital artifact or a physical ID.

Important Note: Anti-Spoofing and Anti-Tampering Checks are disabled for Non-ID Documents, such as Power and Appliance Bills. These types of documents are typically printed on paper and lack certain safeguards, which prevent the FaceTec SDK from making the necessary assessments with a high level of confidence.

ID Scan Anti-Spoofing & "Authenticating" a Photo ID

The ID Scan Digital Spoof Check is ~97% accurate on Photo IDs when tested with random attacks.  

Anti-spoofing for ID Documents will never be as accurate as 3D Liveness, since the physical signal in digital photos of ID Documents is often inconclusive. 

Don't believe us? Have you heard that IDV companies can Scan Photos of ID Documents and determine if they are "Authentic?" Hacker Noon tested many of the major eKYC/IDV providers and found all accepted Fake Documents, so don't believe the hype.

Some ID types will result in a more confident Physical vs. Digital decision when our AI looks to determine that a Physical ID is present. Over time, FaceTec will continue to get better at Photo ID Anti-Spoofing, but it won't ever be foolproof.

Developers should also be aware that physical fake IDs are rampant. On the Dark Web, hyper-realistic Fake IDs can be purchased for $50 and mailed wherever you would like, or sent and used via a proxy IP. They will become cheaper, even more realistic, and more and more pervasive, and there are already millions being imported into many countries each year.

FaceTec's 3D Liveness and Matching are observing much stronger physical signals from a Live 3D Human than any "authenticity" signal from an ID Document.  Because of this, we suggest that Developers look into our 1:N Search for De-duplication. Once a fraudster passes a 3D Liveness Check, even if they have a physical Fake ID in their possession that looks real to FaceTec AI, you can overcome this by scanning the User's 3D FaceMap against all the other 3D FaceMaps of the Users that have onboarded, and no matter what name the used, and if there is a high match, flag it for further analysis.

With a 3D FaceMap Search DB, if someone commits fraud on your platform, their 3D FaceMap can be added to a 3D-DB of 3D FaceMap fraud. Once added, this User will have a very hard time ever getting another account with their face, even under a fake name. This does not prevent the sophisticated attacker willing to create physical fake IDs and use their own face of another person's Live 3D Face from getting that first account, but it works to stop any person from getting multiple accounts under different names.

To prove legal identity remotely, live biometric data must be securely collected from the user and be corroborated by a trusted source, like a Govt ID issuer or by a trusted 3rd-Party who has verified the biometric data matches a Govt ID database (3D FaceMap = Legal ID Portrait Photo), or verified it against photos/FaceMaps in a legal identity issuer's trusted database previously.

Building Your IDV & KYC Process

FaceTec enables you to create Identity Verification (IDV) and Know Your Customer (KYC) Services by providing Free & Unlimited Access to FaceTec's ID Scan Features and Tools.

FaceTec's ID Scan Solution is designed to capture the Document's Data and return a comprehensive set of IDV-related flags to use within Your IDV & KYC System.  Please make sure you design your business logic around the behaviors and flags that FaceTec ID Scan returns.

While there are still several Retry & Rejection Modifiers and Logic Flow Modifiers to customize FaceTec's ID Scan Behavior, these modifiers are only intended to supplement the logic used by Your IDV & KYC System.

When processing a User within Your App, Your IDV & KYC System should make decisions based on all of its available resources, such as the data and flags returned by FaceTec's ID Scan Solution, the Matching Results of the User against a Trusted Data Source, and the results of checking for the User in Your 1:N Bad Actor Database.

ID Anti-Face Tampering & Anti-Photo Swap (Beta)

All ID Scans perform Face Tampering & Anti-Photo Swap checks. FaceTec's ID Scan UI will force the User to Retry once if the Tampering Status cannot be confirmed, but will not block subsequent repeated User Scans.  Optionally, you can Force a Retry until this Succeeds.

To see the result of this check, check the FaceTecFaceOnDocumentStatus on the Match3D2DIDScanResult object. There are 3 possible values:

  1. LikelyOriginalFace – Indicates the Face on the Document has likely not been edited or swapped.
  2. CannotConfirmIDIsAuthentic – When CannotConfirmIDIsAuthentic is returned, either the Face Photo has been modified or replaced, or a higher confidence result could possibly be returned if the User retries and captures better data. Please note, that this does not indicate that FaceTec flagged the ID Scan as a Spoof or Fraud. It could mean the ID Scan was a poor capture, which could be resolved on a User Retry.
  3. NotAvailable – Typically occurs if it is impossible to perform this check on the data provided. For example, if there was no face on the Document side you scanned.
  4. OCRTemplateDoesNotSupportDetection – Indicates the OCR Template is configured to allow the use of taped, glued-on, pasted, or other types of affixed Face Photos over the ID. FaceTec cannot confirm the ID's face photo is the original because this ID is designed to allow for ID Photos that are not actually embedded in the original ID at the time of creation.

The Anti-Face Tampering & Anti-Photo Swap check result is stored by default in the FaceTec Server and shown on the FaceTec Dashboard under the “Photo ID Image Original Face” column.

photo-id-image-original-face-dashboard-col

ID Anti-Text Tampering & Anti-Text Swap (Beta)

All ID Scans perform Text Tampering & Anti-Text Swap checks. FaceTec's ID Scan UI will force the User to Retry once if the Tampering Status cannot be confirmed, but will not block subsequent repeated User Scans.  Optionally, you can Force a Retry until this Succeeds.

To see the result of this check, check the FaceTecTextOnDocumentStatus on the Match3D2DIDScanResult object. There are 3 possible values:

  1. LikelyOriginalText – Indicates the Text on the Document has likely not been edited or swapped.
  2. CannotConfirmIDIsAuthentic – When CannotConfirmIDIsAuthentic is returned, either the Text has been modified or replaced, or a higher confidence result could possibly be returned if the User retries and captures better data. Please note, this result does not indicate FaceTec flagged the ID Scan as a Spoof or Fraud. It could mean the ID Scan was a poor capture, which could be resolved on a User Retry.
  3. NotAvailable – Typically occurs if it is impossible to perform this check on the data provided. For example, if there was no text on the Document side you scanned.

The Anti-Text Tampering & Anti-Text Swap check result is stored by default when using the Facetec Server and is also shown on the FaceTec Dashboard under the “Photo ID Image Original Face” column.

Security Watermark & Hologram Detection (Beta)

FaceTec checks for Security Watermarks and Holograms on the 400+ OCR Templates with Watermark & Hologram Detection Enabled. Watermark & Hologram Detection is enabled whenever a Watermark or Hologram has been marked and specified in the OCR Template Creator. A maximum of two Watermarks/Holograms can be marked on each OCR Template.

ID Scans will be rejected the first time that an Expected Watermark/Hologram is not detected. This Retry Behavior can be ignored by using the IGNORE_WATERMARK Retry Modifier.

The result of this check can be found in the watermarkAndHologramStatus field of either the FaceTecIDScanOnlyResultV10 or FaceTecMatch3D2DIDScanResultV10 object. There are three possible values:

  1. NOT_AVAILABLE – No Watermark or Hologram was configured for the OCR Template, or other prerequisites, like Document Type Detection or Full ID Detection, did not succeed.
  2. DETECTED – The Expected Watermark/Hologram is likely present on the Document.
  3. WAS_NOT_DETECTED – It could not be determined with a High-Degree of Confidence if the Expected Watermark/Hologram is present on the Document. This does not indicate that FaceTec flagged the ID Scan as a Spoof. It could mean the ID Scan was a Poor Capture, which could be resolved on a User Retry.
Important Note: This Feature was designed to detect Watermarks or Holograms on the supported documents and does not perform any Tampering Checks on the detected Watermarks and Holograms printed on the scanned documents.

Checking the Photo ID Match Results

To perform a Photo ID Match, you must start as shown in our Sample Apps using the FaceTec Session constructor that accepts a FaceTecIDScanProcessor. The FaceTecIDScanResult contains the FaceTecIDScan object, which must be passed into the FaceTec Server for evaluation.

On the result object returned by the FaceTec Server, check the FaceTecFullIDStatus to check whether FaceTec assessed the user scanned their Full ID. Note: The Match Level reported by the Photo ID Match will ALWAYS be zero if the FaceTecFullIDStatus is FullIDNotDetected.

On the result object returned by the FaceTec Server, check the FaceTecDigitalSpoofStatus to determine whether the scan was attempted against an ID that is presented on a digital screen. 

On the result object returned by the FaceTec Server, check the FaceTecFaceOnDocumentStatus to determine whether the scan's face photo may have been modified or replaced. Note that a LikelyOriginalFace result is not required for the scan to succeed. However, you may force a retry in the FaceTec Server based on your own success criteria.

On the result object returned by the FaceTec Server, check the FaceTecTextOnDocumentStatus to determine whether the scan's text may have been modified or replaced. Note that a LikelyOriginalText result is not required for the scan to succeed. However, you may force a retry in the FaceTec Server based on your own success criteria.

On the result object returned by the FaceTec Server, check the FaceTecWatermarkAndHologramStatus to determine whether the ID text may have been modified to remove, cover, or alter its Security Watermarks and or Holograms. Note that a Detected result is not required for the scan to succeed. However, you may force a retry in the FaceTec Server based on your own success criteria.

FaceTec Server Dashboard Display

These results also appear when viewing the FaceTec Server Dashboard.