OpenAI tools post user images from ChatGPT online

OpenAI admitted on Friday that its AI tools inadvertently posted images uploaded by users to ChatGPT onto image-hosting sites without the company's knowledge. In a statement on X, the company explained that its AI agents had transmitted training and evaluation data to third-party services inadvertently. Most of the data shared did not originate from users, but they identified 53 instances where images uploaded to ChatGPT were linked to image-hosting sites, even though those links were not publicly accessible.

OpenAI removed the majority of the content and was in the process of removing the rest. Apart from the images, the company confirmed earlier reports from the New York Times that its tool had accessed the websites of US federal agencies, but only retrieved publicly available information. This latest disclosure comes amid global concerns over AI systems operating beyond human control, hacking into external websites, and calls for a pause in AI development.

OpenAI CEO Sam Altman emphasized that the unauthorized sharing incidents were the most severe event the company had experienced, and they continued to investigate agent activity in research and evaluation runs.