An artificial intelligence company has deleted approximately 3 million user photos sourced from the dating platform OkCupid following a US regulatory investigation into data-sharing practices.
The issue stems from a 2014 data transfer in which OkCupid provided user images, along with demographic and location data, to Clarifai, a US-based artificial intelligence company that develops facial recognition systems. According to the US Federal Trade Commission, the data sharing occurred without informing users and was inconsistent with the platform’s stated privacy policies.
The FTC opened an investigation into the arrangement after reports surfaced in 2019. In March 2026, the agency reached a settlement with OkCupid and its parent company, Match Group, over allegations that users had been misled about how their personal data would be handled.
As part of the outcome, Clarifai confirmed that it had deleted both the dataset and any machine learning models trained using the images. The company certified the deletion to regulators in April 2026 and stated that the data had not been shared with third parties.
The photos were used to train facial recognition systems capable of identifying attributes such as age, gender, and other characteristics. These systems rely on large image datasets to improve accuracy and performance.
Regulators said the original data transfer violated federal rules on deceptive business practices, as OkCupid’s privacy policy indicated that user information would not be shared with unrelated third parties without consent.
The settlement does not include financial penalties, as the FTC does not have the authority to impose fines for this type of violation under current law. However, the agreement requires OkCupid and Match Group to avoid misrepresenting their data practices in the future and to comply with ongoing oversight conditions.
Clarifai was not accused of wrongdoing in the case. The company stated that it had responded to regulatory concerns by removing the data and associated models after the settlement was reached.
The investigation remains part of broader scrutiny into how user-generated data is used to train artificial intelligence systems. Authorities have not announced additional enforcement actions related to the case.