FTC finalises order against IntelliVision over “bias-free” facial recognition claims

On 13 January 2025 the U.S. Federal Trade Commission announced that it had finalised an order against IntelliVision Technologies Corp., settling allegations that the San Jose company made false, misleading, or unsubstantiated claims about its AI-powered facial recognition software — including that the software was free of gender or racial bias and had one of the highest accuracy rates on the market.

The Commission’s complaint, first announced in December 2024 and issued with the final order materials in January 2025 (Matter 232 3023; Docket C-4809), describes a product sold to original equipment manufacturers, integrators, and large end users, and integrated into consumer devices such as the 2GIG Edge home security system and the Elan Intelligent Touch Panel. From late 2018 through early 2024, IntelliVision’s website, intelli-vision.com, carried claims that the software could “detect faces of all ethnicities, without racial bias,” offered “Zero gender or racial bias through model training with millions of faces from datasets from around the world,” and that anti-spoofing technology “ensures the system cannot be fooled by a photo or video image.” In presentations to trade customers, the company claimed one of the highest accuracy rates on the market.

Those assertions are the part of the record most summaries compress. The complaint alleges IntelliVision did not possess testing to support the highest-accuracy claim, the “without racial bias” claim, or “zero gender or racial bias.” It further alleges the company did not train the software on millions of faces; rather, it trained on images of approximately 100,000 unique individuals and then used technology to create multiple variants of those same images. IntelliVision had submitted algorithms to the National Institute of Standards and Technology for testing in 2019, 2022, and 2023; NIST’s published results, the complaint states, showed error rates that differed across demographics including region of birth and sex, and showed the algorithms were not among the top-performing systems — for example, as of 19 December 2023, not among the top 100 by false non-match rate. The complaint also alleges IntelliVision lacked sufficient testing for its unqualified anti-spoofing claim and did not assess how anti-spoofing performed across demographic groups.

The final Decision and Order, issued 8 January 2025, records that IntelliVision neither admits nor denies the complaint’s allegations except as needed for jurisdiction. It prohibits misrepresentations about accuracy or efficacy; about comparative performance across genders, ethnicities, and skin tones; and about spoofing or liveness detection. Separately, it bars representations about effectiveness, accuracy, lack of bias, or spoofing detection unless the company possesses and relies on competent and reliable testing at the time of the claim, with documentation covering methods, image sources, demographics and skin tone where collected, and results. After receiving no public comments, the Commission voted 5–0 to approve the final order.

The pattern is expert-it-isn’t: marketing treated demographic fairness and top-tier accuracy as settled product features. The enforcement action treats them as claims that require competent, documented testing — the constraint the system’s sales copy had left out.