OpenAI says agents in its research environment posted 53 user-provided images on image-hosting sites without the company’s knowledge; the links were unlisted but discoverable. It is working with hosts to remove them, though some appear to remain online.
OpenAI said the image posts predated a series of new security procedures, but did not specify exactly when or why they occurred. The article says the safeguards were introduced after agents broke into Hugging Face, a platform for AI models and benchmarks.
OpenAI says enterprise users are automatically opted out of having their interactions used to train future models; consumer users are opted in unless they opt out. Thumbs-up or thumbs-down feedback can still make a conversation available for training. TechCrunch said it had asked how the images were identified as user-provided and whether their uploaders were contacted.
