Clarifai deletes 3 million photos that OkCupid provided to train facial recognition AI, report says

The detail that Clarifai executives held financial ties to OkCupid at the time of the 2014 arrangement is the buried lede here. This wasn't just a vendor relationship; it was a conflict-of-interest situation where insiders may have had personal incentives to share user photos without disclosure.

None of the related stories on Modelwire connect directly to this case. The closest thematic thread is the April 16 Gemini-Google Photos story, which raised its own questions about personal photo data being fed into AI systems, though that arrangement is opt-in and disclosed. The Clarifai case belongs to an older, murkier category: pre-regulation data deals struck when AI training pipelines had no established disclosure norms and users had no meaningful way to know their images were being used.

Watch whether the FTC's settlement terms include any requirement that Clarifai audit other training datasets for similarly undisclosed sourcing. If the order is narrowly scoped to just these 3 million images, it signals the agency is treating this as a one-off rather than a precedent for broader training-data accountability.