Clarifai Deletes 3M OkCupid Photos Used for AI Training

The AI platform Clarifai has permanently erased a dataset of three million user photos it obtained from the dating service OkCupid, a move following a Federal Trade Commission investigation. According to reports, the company also deleted any AI models trained on that data, which was originally gathered to develop facial recognition technology. This case highlights significant concerns over user consent and data privacy, as the sharing of these photos reportedly violated OkCupid’s own stated privacy policies at the time.

The data transfer originated in 2014, when Clarifai, which had received investment from OkCupid executives, requested access to the dating app’s information. OkCupid subsequently provided millions of user-uploaded photographs alongside other demographic and location details. Internal communications reveal the intent behind the request, with Clarifai’s CEO noting in an email the platform’s “HUGE amount of awesome data” for AI training purposes. Despite occurring over a decade ago, the incident only drew regulatory scrutiny years later, after a 2019 news report detailed how Clarifai used the images to build a tool for estimating age, sex, and race from facial features.

Last month, the FTC and OkCupid’s parent company, Match Group, reached a settlement in the resulting lawsuit. While neither OkCupid nor Match admitted to allegations of deceiving users, Clarifai’s confirmation of the data deletion strongly implies the photos were improperly accessed. The FTC further alleged that, since 2014, the companies deliberately concealed the data sharing and attempted to obstruct the investigation. As part of the settlement, OkCupid and Match are now permanently prohibited from misrepresenting their data collection and sharing practices, a prohibition that reinforces existing FTC rules. Although the agency cannot levy fines for this type of first offense, the binding order sets a clear precedent for enforcing privacy policy compliance.