'Zero Gender or Racial Bias,' the Company Claimed. The FTC Says There Was No Evidence for It.

The Federal Trade Commission finalized an order against IntelliVision Technologies Corp., settling allegations that the company made false, misleading, or unsubstantiated claims that its AI-powered facial recognition software was free of gender or racial bias and among the most accurate on the market.DOCUMENTED

San Jose-based IntelliVision's facial recognition software is embedded in millions of consumer devices, including smart home security systems and touch panels sold through partners such as Nice North America.DOCUMENTED

The claims and the missing evidence

According to the FTC's complaint, from late 2018 through early 2024, IntelliVision made numerous claims that its software had "one of the highest accuracy rates on the market" and performed with "zero gender or racial bias" through model training on "millions of faces" from datasets spanning the globe.DOCUMENTED The company also claimed its anti-spoofing technology could not be fooled by photos or video images presented to the camera.DOCUMENTED

"Companies shouldn't be touting bias-free artificial intelligence systems unless they can back those claims up," said Samuel Levine, then-Director of the FTC's Bureau of Consumer Protection. "Those who develop and use AI systems are not exempt from basic deceptive advertising principles."DOCUMENTED According to the FTC, IntelliVision did not have data to support its claims that the software had one of the highest accuracy rates on the market or performed with zero gender or racial bias, and the company had not trained the software the way its marketing described.DOCUMENTED

Why the bias claim mattered specifically

Facial recognition technology has faced sustained, well-documented scrutiny over its history of uneven accuracy across different demographic groups, with independent research repeatedly finding higher error rates for women and people with darker skin tones in many commercial systems.REVIEWED A company claiming its system had solved that problem entirely — "zero" bias, not merely reduced bias — was making an unusually strong and specific claim, one the FTC's complaint says was not backed by the kind of rigorous, demographic-stratified testing that would be needed to support it.REVIEWED

What the final order requires

The final order bars IntelliVision from making unsubstantiated statements about its technology's accuracy or efficacy, requires the company to support any such claims with "competent and reliable testing," and imposes recordkeeping obligations so the FTC can verify compliance over time.DOCUMENTED The order runs for 20 years, a duration reflecting the FTC's typical approach for consent agreements addressing ongoing product-marketing practices rather than a one-time transaction.DOCUMENTED

The company claimed "zero gender or racial bias." The FTC says it never had the data to back that up.

Why the case is broader than one vendor

IntelliVision's technology reaches consumers indirectly, embedded inside other companies' branded security and smart-home products, meaning many people who rely on facial-recognition-equipped devices in their homes or businesses may not know IntelliVision's software is the underlying engine, let alone that its bias and accuracy claims were unsubstantiated.REVIEWED The FTC's business guidance following the settlement urged any company selling or deploying facial recognition or other biometric identification technology to ensure its own marketing claims about accuracy and fairness are backed by testing — a reminder aimed as much at IntelliVision's downstream commercial partners and customers as at the company itself.REVIEWED

The case adds to a growing body of FTC enforcement specifically targeting AI accuracy and bias claims across industries, from facial recognition to health and financial technology, reflecting the agency's consistent position that AI-specific marketing claims are held to the same evidentiary standard as any other product performance claim.REVIEWED

The compliance burden the order creates

Beyond barring specific unsupported claims, the 20-year consent order requires IntelliVision to maintain detailed records of whatever testing it conducts to support any future accuracy or bias-related marketing, giving the FTC a mechanism to audit the company's claims well beyond the immediate settlement.REVIEWED That kind of extended recordkeeping obligation is a common feature of FTC settlements involving technical performance claims that are difficult for an outside party, including a regulator, to verify without access to a company's internal testing data — the order effectively shifts the burden onto IntelliVision to proactively document and preserve the evidence behind any future marketing statement, rather than leaving enforcement dependent on the FTC independently reconstructing that evidence after the fact.REVIEWED

For businesses and consumers relying on any AI-branded security product, the IntelliVision case is a reminder that a vendor's marketing claims about accuracy, fairness, or bias should be treated as claims requiring independent verification, not assumed true simply because the underlying technology carries an AI label.REVIEWED

The Commission's 5-0 vote to accept the settlement, together with a concurring statement from Commissioner Andrew Ferguson, reflects bipartisan agreement within the agency that unsupported AI-bias claims warrant enforcement regardless of broader disagreements commissioners may have on other AI-policy questions.DOCUMENTED That consensus stands in contrast to the split vote in the FTC's Rytr case the same year, suggesting the specific legal theory matters: a straightforward false-advertising claim about a product's tested accuracy draws more consistent agreement across the Commission than a broader "means and instrumentalities" theory applied to a general-purpose writing tool.REVIEWED

Because IntelliVision's software operates largely behind the scenes of other companies' branded products, consumers who want to know whether a given smart-home security device relies on IntelliVision's technology, and is therefore covered by this settlement's marketing restrictions, may need to check with the device manufacturer directly rather than assuming any particular product is or isn't affected.REVIEWED The settlement does not require IntelliVision's commercial partners to notify their own end customers of the order, leaving that gap in public awareness as an open question for however long the 20-year consent agreement remains in effect.REVIEWED

Have documents relevant to this story? Reach us through our tips channel.