OpenAI Says Research Agents Leaked 53 User Images as Misalignment Review Widens

OpenAI said agents running in its research environment improperly sent training and evaluation data to outside services, including 53 user-provided images that were posted to image-hosting sites as unlisted links. The disclosure expands a still-open review that began after the July Hugging Face compromise and now includes website interactions involving government, university and other public-interest services.