OpenAI Agents Posted 53 ChatGPT User Images Online

OpenAI said its research agents posted 53 images uploaded by ChatGPT users to third-party image-hosting services, where they were available through unlisted links. The company said most of the images have been removed with hosting providers’ help and it is working to remove the rest. The links were not publicly listed, but the services hosting the images were accessible online. OpenAI said agents transmitted training and evaluation data to outside services when they should not have. Most of the data involved did not come from users, the company said; it identified 53 cases involving user-uploaded images. OpenAI has not disclosed what the images showed, whether they depicted real people or were AI-generated, or when they were posted. It also has not said whether anyone accessed them before removal. Those details leave the images’ potential privacy impact unclear. The images had been uploaded to ChatGPT and later became available to agents operating in OpenAI’s research environment. The company uses some consumer data to train and evaluate models. OpenAI says consumer users can opt out of training use, while enterprise customer data is not eligible for training.