Clarifai Erases 3 Million OkCupid Photos from Facial AI Training

The artificial intelligence platform Clarifai has announced the deletion of approximately 3 million images, which it states were obtained from the dating application OkCupid for the purpose of training its facial recognition AI. According to Reuters, the company also purged all AI models that had been developed using this specific dataset.

An investigation by the Federal Trade Commission (FTC) revealed that in 2014, Clarifai requested data from OkCupid, a company whose executives had previously invested in Clarifai. Reports indicate that the dating app subsequently provided these user-uploaded photographs, alongside other demographic and location data. Such data sharing, according to OkCupid’s own privacy policies, should have been prohibited.

“We’re ⁠collecting data now and just realized that OKCupid must have a HUGE amount of awesome data for this,” Clarifai founder and CEO Matthew Zeiler wrote in an email to OkCupid co-founder Maxwell Krohn, as detailed in court documents reviewed by Reuters.

Despite the initial data sharing incident reportedly occurring a decade ago, the FTC did not commence an investigation until 2019. This action was prompted by a New York Times article that brought to light Clarifai's use of images from OkCupid to construct an AI tool capable of estimating an individual’s age, sex, and race based on their face.

The FTC and OkCupid, which is owned by Match Group, reached a settlement in their lawsuit last month. While OkCupid and Match Group did not admit to allegations of deceiving users by violating their own privacy policies, Clarifai’s confirmation of data deletion implicitly supports that the company did indeed gain access to those photos. The FTC further alleged that since 2014, Match Group and OkCupid deliberately concealed this behavior and attempted to obstruct its investigation.

Neither OkCupid nor Clarifai immediately responded to TechCrunch’s requests for comment on the matter.

Although the FTC is not empowered to levy fines against companies for this particular type of first-time offense, the agency declared that OkCupid and Match are "permanently prohibited from misrepresenting or assisting others in misrepresenting" the nature of their data collection and sharing. This ruling thus reinforces existing prohibitions against such behaviors, which are already deemed impermissible by the FTC.