FB pixel

IntelliVision censured for misleading biometric accuracy and bias claims by FTC

Claims undermined by NIST test results
IntelliVision censured for misleading biometric accuracy and bias claims by FTC
 

The U.S. Federal Trade Commission has slapped IntelliVision with a consent order to halt claims about the accuracy of its face biometrics and liveness detection software than it cannot adequately support.

IntelliVision’s claims that its facial recognition is free from gender and racial bias and is among the most accurate on the market were not backed up by evidence, according to the FTC. Further, the company’s claims to have trained its biometric algorithms on millions of faces were found to be false, IntelliVision having really used images of approximately 100,000 people “and then used technology to create variants of those same images.”

The company’s presentation attack detection technology was advertised as ensuring protection against spoofing with photos or videos, which was also found to lack support.

IntelliVision is now banned from making any claims about the accuracy, lack of bias or spoof protection unless it can be backed by “competent and reliable testing.”

The problem is not that such testing has not been performed.  It is that testing by NIST shows matching rates that vary between different demographics, and does not show a false non-match rate within the top 100 for the last software submitted by the company to the FRTE 1:1, in December, 2023. IntelliVision had also submitted an algorithm to NIST’s FRTE 1:N in 2022, which currently sits 74th in Border ΔT ≥ 10 YRS and 95th in the Visa-Kiosk category.

The FTC observed the discrepancies between claims and fact beginning in late-2018, soon after the company was acquired by Nortek Security & Control. IntelliVision’s facial recognition is used in smart home systems such as panels sold by Nice North America, the FTC notes.

The Commission voted unanimously to issue the complaint and accept the company’s settlement agreement. Each violation of the agreement will incur a civil penalty of up to $51,744.

A biometrics policy statement from the FTC last year focussed mainly on data protection issues, but action against Rite Aid settled later in the year touched on both data protection and the software’s effectiveness.

The Commission has also published a blog post with tips for biometrics providers on how “to avoid breaking the law and misleading consumers.”

Related Posts

Article Topics

 |   |   |   |   |   |   |   |   | 

Latest Biometrics News

 

Liquid identity verifications surge past 60M as Japan leans into chip-scanning

Liquid has reached the 60 million digital identity verification milestone with its online KYC service, with a surge in verifications…

 

Car dealerships rev up digital ID verification to counter rise in identity fraud

Whether it’s a fake credit history, a phony license or a test driver with a stolen identity who makes tracks…

 

GovTech to deliver $10 trillion in value by 2034, says WEF

At the meeting of the World Economic Forum (WEF) in Davos this week, tech is front and center – and…

 

Davos discusses digital wallets, AI economy

This year’s Davos World Economic Forum (WEF) is bringing not only tense trade talks between the U.S. and Europe but…

 

ASEAN updates guidance on deepfakes

The threat of deepfakes is entering high-level discussions from Southeast Asia to Davos. The Association of Southeast Asian Nations (ASEAN)…

 

Philippines faces 36 million backlog in ID cards

The Philippines are still facing a 36 million backlog in distributing the country’s national ID cards which will need additional…

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Viewed This Week

Featured Company

Biometrics Insight, Opinion

Digital ID In-Depth

Biometrics White Papers

Biometrics Events