What privacy issue led to Clarifai deletion?

Clarifai deleted millions of OkCupid photos after FTC settlement

Clarifai said it deleted about 3 million photos that it received from OkCupid to train facial-recognition models, along with the associated training models. The move follows a U.S. FTC settlement with OkCupid over privacy violations.

The key compliance timeline centers on how the data was obtained and used. Clarifai received the photos in 2014, and the materials were transferred without users’ knowledge, according to the reporting summarized in the story corpus. That transfer reportedly conflicted with OkCupid’s own privacy policy, and regulators later pursued the privacy and consent failures.

Why this matters

Facial recognition model training depends on large-scale datasets, and this case highlights that the dataset provenance—whether consent was obtained and whether the original promise to users was honored—can become a regulatory and operational risk.

It also underscores how FTC actions can extend beyond the original company named in the settlement. Even though Clarifai is the AI developer doing the model training, it became the entity expected to remove the outputs of the underlying data pipeline.

What Clarifai did

• Removed 3M photos used for training
• Deleted the models trained on those images

No detailed technical description was provided about the architecture of the models or whether any additional retraining happened, but the deletion signals a remediation step tied directly to the FTC case.

Overall, the episode is another example of how AI companies face downstream obligations when privacy violations occur upstream in data sharing and partnerships.