authID.ai is celebrating a major milestone after passing the first and second level of the ISO 30107-3 evaluation. The test is a measure of liveness detection, and gauges a solution’s ability to identify the use of spoofs in the authentication process.
The actual examination was carried out by iBeta, using authID.ai’s flagship Verified solution. Verified was able to pass both Level 1 and Level 2 with a perfect score, successfully spotting all 360 spoofing attempts at each level (for a total of 720) to achieve a failure rate of zero percent.
In plain terms, that means that Verified is very good at distinguishing real faces from the photos, masks, or digital images that a fraudster might use to try to get past authentication software. authID.ai believes that its offering can help mitigate the threat of fraud, noting that the price tag for digital identity fraud now exceeds $43 billion on an annual basis.
Verified uses both active and passive liveness detection techniques to perform Presentation Attack Detection. With regards to the former, users are asked to smile when submitting a selfie to prove that they are present and are opting into the authentication process. The passive component, on the other hand, analyzes a series of images to gauge their authenticity.
“This independent confirmation demonstrates how Verified can provide additional fraud protection and security to authID customers,” said authID.ai Product SVP Jeremiah Mason. “Our ISO 30107-3 Level 2 compliant PAD system prevent presentation attacks and provide stronger safeguards against fraud and account takeover by authenticating the person instead of what they know or have.”
authID.ai released an updated version of Verified at the beginning of the year. The solution is designed primarily for remote onboarding, and presumably helped the company secure $22.5 million in investment funding back in March. Verified is not the first solution to pass the iBeta evaluation with a perfect score, with other solutions like CyberLink’s FaceMe engine also accomplishing the feat.
(Originally posted on FindBiometrics)