Regulatory Agency Alleges Deceptive Marketing Practices by AI Video Company

Topics: Artificial Intelligence & Machine Learning,
Government,
Industry Specific

IntelliVision Agrees to Settlement with FTC Over Misleading Facial Recognition Claims

Regulator Accuses AI Video Firm of Deceptive Marketing
IntelliVision’s AI-driven facial recognition software is integrated into millions of smart devices. (Image: Shutterstock)

Facial recognition technology firm IntelliVision has settled with the Federal Trade Commission (FTC) following allegations of misleading marketing practices. Key claims made by the company, including assertions that its technology offered “zero gender or racial bias” and was market-leading in accuracy, were deemed false by regulators.

The FTC detailed in a recent complaint that from late 2018 through early 2024, IntelliVision propagated numerous misleading statements. Notably, the company claimed its software could accurately identify faces of all ethnicities without bias and relied on a dataset of “millions of images from across the world” for training.

The commission unanimously voted to formalize the complaint and approved a settlement with the California-based company, which should not be conflated with Intellivision, a gaming console introduced by Mattel Electronics in 1979.

Sam Levine, director of the FTC’s Bureau of Consumer Protection, stated, “Companies must not promote bias-free AI systems without substantiating those claims.” He emphasized that AI developers and users are subject to standard advertising regulations, highlighting the necessity for accountability in the marketing of such technologies.

IntelliVision claims its facial recognition software is deployed in over 15 million devices worldwide, particularly in residential security systems and smart home applications. Unlike competitors such as Ring and Arlo, IntelliVision primarily sells its technology to original equipment manufacturers and large-scale integrators rather than directly to consumers. Its products include critical components for home security systems, empowering users to utilize biometric authentication.

However, the FTC’s complaint identified that IntelliVision made false assertions regarding the presence of anti-spoofing “liveness tests” within its software, which purportedly prevent the use of pre-recorded videos or static images to deceive the system. The firm also claimed its technology boasted a leading accuracy rate, a statement that the FTC found was not adequately supported.

The FTC indicated that IntelliVision’s training dataset was not as extensive as claimed; in reality, the company utilized images from approximately 100,000 unique individuals and augmented those images for its models. Testing conducted by the National Institute of Standards and Technology (NIST) revealed that IntelliVision’s performance varied significantly based on a person’s region of origin and sex, a common trend within camera-based recognition systems.

In 2019 and the following years, IntelliVision’s algorithms were submitted for examination by NIST, which uncovered significant discrepancies in performance, notably revealing higher false positive rates for certain demographic groups. By December 2023, IntelliVision’s algorithms did not rank among the top performers in reducing false non-matches, further undermining the company’s claims of superiority.

IntelliVision has not publicly commented on the allegations or the settlement terms, which commonly entail neither admitting nor denying the FTC’s claims. Should any violation of the settlement occur, the agency retains the right to pursue legal action. This case is the second notable facial recognition-related settlement for the FTC, which previously prohibited Rite Aid from utilizing such technology for surveillance due to inadequate checks against the risks of inaccuracies linked to race and gender.

In a concurrent statement, FTC Commissioner Andrew N. Ferguson noted that the agency does not assert that all AI systems are devoid of bias. He highlighted the need for clarity in IntelliVision’s marketing, suggesting that their references to “bias” needed further definition to avoid consumer confusion.

Recently, IntelliVision underwent ownership change, transitioning from Nice North America to Nipun Vision, the latter founded by Vaidhi Nathan, former CEO of IntelliVision, now serving as chairman of the new entity. IntelliVision intends to redirect its focus towards security, surveillance, and advanced driver assistance systems (ADAS), indicative of its evolving business strategy.

Source link