Better fraud detection fights complex ID document fraud, says Yoti report

It’s been widely noted in the biometrics space that fraud has evolved with the rise of deepfakes, generative AI, and more intricate document tampering, but also better monitoring software. And now, fraud rates are trending upwards because fraud detection is improving, according to the first edition of the Identity Fraud Report from identity verification provider Yoti.

No one actually knows for sure to what extent fraud is increasing or decreasing. Data on fraud can only go based on businesses reporting cases. And now, monitoring software can detect suspicious activity, AI and automation can detect fraudulent identity documents, and external and third party databases are being better leveraged.

Around two thirds of the fraud attempts Yoti sees are impersonation attempts, where a user submits a selfie that doesn’t correspond to the identity document at hand. The other third consists of fraudulent documents.

Documents might be tampered with, counterfeit, or a sample document that’s made publicly available. An attacker might also use a picture of someone else’s genuine document that might be found online. A document might also be completely fabricated and are sometimes called “fantasy” documents.

Document fraud varies from low quality, such as fantasy documents, which can take up to 5 to 30 minutes to create and cost less than £50 (roughly US$63), to highly sophisticated fraud involving fraudulently obtained genuine documents. These can take over 4 hours and £1000 ($1,263) to complete and require a third party in order to confirm that the document is fraudulent.

Half of all document fraud that Yoti sees involves driver’s license and national IDs.

Online resources and tech advancements make the creation of fake IDs much easier. Sites like OnlyFake will make a synthetic ID supposedly through the use of AI for $15, while others will provide security features like replicas of holograms in bulk. One site sells 1,000 for less than £200 ($253).

Yoti uses a combination of automated authenticity checks and humans conducting verification and counter fraud. Authenticity checks determine that a document is legitimate and hasn’t been tampered with. Yoti’s MyFace biometric verification compares an individual’s face to that on an ID. Liveness detection can identify presentation attacks like deepfake videos, paper images, and screen images. The company achieved a 100 percent detection rate for its liveness detection algorithm in iBeta testing.

Those who fail the automated biometric check will undergo a human fallback check. If a user doesn’t understand the instructions or is taking a photo in poor lighting, they might need to undergo a human check. A user may also undergo the check if they submit an identity document that was taken years ago when their appearance was very different. As many as 30 percent of verifications may need to be reviewed by a human team.

Article Topics

biometric liveness detection  |  biometrics  |  document verification  |  fraud prevention  |  Yoti