Clarifai Removes 3M Photos from OkCupid in AI Training Data Cleanup

Clarifai has deleted 3 million photos that dating platform OkCupid provided for training its facial recognition AI system, following an FTC settlement. The data sharing arrangement, which began in 2014 with involvement of OkCupid executives who had invested in Clarifai, has now been unwound as part of regulatory compliance.

Clarifai, an artificial intelligence company specializing in computer vision and image recognition, has removed approximately 3 million photographs from its systems that originated from OkCupid, the online dating platform. The deletion was completed as part of a settlement agreement with the Federal Trade Commission regarding the company's use of personal data for training facial recognition models.

According to court documents, the initial data sharing arrangement between Clarifai and OkCupid began in 2014. The agreement appeared to involve financial connections, as executives from OkCupid had made investments in Clarifai at that time. OkCupid provided the dating platform's user photos to help train and improve Clarifai's artificial intelligence systems for facial recognition capabilities.

The Federal Trade Commission's settlement with Clarifai represents part of a broader regulatory push against companies that collect and utilize personal data without proper consent from individuals. The removal of millions of photos underscores growing concerns about how tech companies handle sensitive user information, particularly when that data is repurposed for artificial intelligence training without explicit permission from the original subjects.

This case highlights tensions within the AI industry between the need for large datasets to train sophisticated machine learning models and the privacy rights of individuals whose images may be included in those datasets. The settlement signals that regulators expect companies to be more transparent about data usage and to obtain appropriate consent before deploying personal photographs for AI training purposes.