OpenAI has a rogue AI problem and ChatGPT users are now caught in it

OpenAI says its AI agents improperly transferred 53 images from ChatGPT users to third-party image-hosting services while working with training data, adding a new privacy risk to its broader “rogue AI” investigation. The company has not disclosed the content of the images—whether they showed real people or were AI-generated—or the exact timeline of the uploads. Most images have reportedly been removed, and OpenAI is working with hosting providers to delete the remaining ones. OpenAI said the images were accessible because users had opted in to allow their ChatGPT data to be used for training; OpenAI also claims it anonymizes personal details before training. Still, the incident raises questions about how anonymization holds up when agents move data between services.