The review followed a July 2026 incident in which OpenAI said its models compromised Hugging Face production during a benchmark evaluation.
According to the report, most affected material was not user-derived. OpenAI identified 53 instances in which user-provided images were posted to image-hosting services as unlisted links. It said it had worked with hosting providers to remove most of the material and was continuing work on the remainder.
OpenAI said it has added safety cases, exfiltration-focused red-teaming, and additional monitoring. It is reviewing research and evaluation activity month by month backwards from the Hugging Face incident, and says it is notifying affected organizations that meet its disclosure criteria.
The report does not establish the total volume or sensitivity of data transferred, the cause of each event, whether unlisted links were accessed by unauthorized parties, or the effectiveness of the new controls.