The Federal Trade Commission (FTC) on Tuesday announced an enforcement action against a facial recognition technology company for allegedly deceptively marketing its software as being accurate and free of gender and racial bias.

The company, IntelliVision Technologies, has asserted that its facial recognition software is one of the most accurate available and that it “performs with zero gender or racial bias,” but had no evidence to support the claims, according to the FTC.

The company’s software is primarily deployed in home security systems and what the FTC called smart home touch panels.

IntelliVision also lied when it said it trained its software on millions of faces, the FTC said. In reality, the company instead trained the product on images belonging to only about 100,000 people. It then created “variants” of those images to further train the product, the agency said.

The company also lacked enough evidence to back up its assertion that its anti-spoofing technology guaranteed its software can’t be deceived by photo or video images, according to the FTC.

A proposed settlement with the FTC prohibits the company from deceiving consumers about several aspects of its product’s performance, including the accuracy and reliability of its software, how it performs across racial and gender groups and how well it identifies spoofing. 

In order to make qualitative claims in the future, IntelliVision must effectively and comprehensively test its system, according to the agency.

The IntelliVision case is just the second FTC action involving facial recognition technology.  Last December, the agency banned RiteAid from deploying facial recognition technology to surveil visitors to its stores for five years. 

The enforcement action was brought in response to consumer complaints that the drugstore’s program unfairly discriminated against minorities due to rampant inaccuracies and led staff to unfairly throw people out of hundreds of stores.