ID R&D innovates again, this time with document liveness detection for identity verification
Document-based presentation attacks against remote onboarding systems have been detected by digital service providers. Document spoofs are a growing trend in identity fraud, prompting the need for a new approach to detect and prevent it.
ID R&D has been proactively addressing emerging challenges in biometrics and identity verification for years, CEO Alexey Khitrov tells Biometric Update, applying its focus on research and expertise in AI to being ahead of the curve with innovative solutions. One example is its frictionless, single-image approach to facial presentation attack detection (PAD).
“In security, there are no silver bullets–you have to fill all the potential gaps without making it painful for legitimate users,” says Khitrov. “At ID R&D we like to foresee those gaps and then leverage AI to address them…without adding friction.”
An urgent problem
One gap that ID R&D saw coming is document spoofing in digital onboarding. In 2021, they provided its first version of a universal document liveness detection algorithm to early adopters, and several of them found the problem to be worse than they expected, according to Khitrov. Fraud analysis after the fact was turning up accounts bound to ID documents that had passed document authentication but later proved to be fraudulent. The software revealed that many documents presented as proof of identity were in fact not photos of physical documents in the possession of the person being onboarded, but rather of documents displayed on a screen. In many cases, they were digitally manipulated to change the portrait or biographic data. The fraudster who successfully completes onboarding can be very difficult to detect and prevent from converting the attack into theft.
ID document authentication traditionally takes place in person, where the check can be carried out by a human with full assurance that the physical document in question is present. But a remote process demands a digital approach.
“Digital onboarding is so critical to making banking and other services more accessible, particularly for the billions of people without a convenient retail option nearby,” commented David Benini, VP of Marketing at ID R&D. “But getting legitimate customers through the process without opening the floodgates to fraud relies on the thorough identity proofing that government-issued identity documents represent.”
Physical ID documents command high prices on the dark web, but scans and photos of ID cards and passports are also plentiful, typically less expensive than physical documents, and often sold with an accompanying selfie.
It follows that when document spoofs aren’t reliably preventable, there’s a risk of scalable fraud. The bad actor can create many fake accounts at a time, Khitrov points out. Unlike traditional physical document manipulation, these attacks are easier to repeat because the manipulations are digital. Any number of ID documents purchased on the dark web can be used, and digital manipulation processes can be automated for speed and repeated attempts.
What is document liveness detection?
ID R&D sees document liveness as a useful AI-powered signal that provides a strong indicator of either legitimacy or fraud. It should be considered an essential part of remote identity verification and complementary to facial liveness detection, according to Khitrov.
Document liveness detection brings the same concepts as are used in face liveness detection to the very first stage in the onboarding process, answering the question: are you proving your identity with live, present evidence, or with a digital replica created to exploit the remote process? Document spoof checks add yet another layer of security to make fraud that much more difficult and risky for bad guys.
How to implement it
ID R&D’s document liveness detection product uses the same single-image, passive approach as its facial liveness detection, which makes its integration as another signal detector quite straightforward. It can be combined with facial liveness detection and biometric comparison of the facial images on the document and the selfie as part of a complete workflow.
ID R&D recently made its document liveness product IDLive Doc generally available, and today announced that it is already being used by several customers around the globe. Because it is not tied to any specific type of identity documents or features, it can be used with passports, driver’s licenses, and national IDs in any country. Like the company’s other technologies, it is designed as a developer-friendly, white-labeled solution to be built into customized applications.
According to the press release, customers are responding well to the product. “ID R&D’s IDLive Doc saved RealAML years of in-house development, helped acquire new customers, and immediately demonstrated its ability to stop fraud attacks,” commented Jordan McCown, CEO and co-founder of RealAML. “IDLive Doc integrated seamlessly into our existing KYC process without disrupting other core components.”
The company’s frictionless approach to liveness detection makes it easy for end-users, and its developer-friendly approach makes it easy for businesses and other organizations to build into their own workflow. Like liveness detection for face, ID R&D expects document liveness to be widely adopted to improve the document checks that are the foundation of trusted identity claims in the online world. More information about document liveness detection can be found at ID R&D’s blog, and the product is described in more depth on their website.