It Took 60+ Days to Remove a CSAM Account on Sora
Yesterday, in a surprise announcement, OpenAI said it would shut down Sora. While the decision has been attributed to the company’s growing focus on enterprise solutions, it is difficult to believe content moderation challenges did not play a role.
For example, on October 15, 2025, one of our researchers uncovered an account on Sora that appeared to be producing what could be classified as child sexual abuse material (CSAM).
The discovery occurred while the researcher was algorithmically served sexually suggestive adult content that appeared to be nonconsensual. After clicking through to the creator’s profile, the researcher was presented with content featuring what appeared to be underage individuals in sexually suggestive videos.
We reported the account to OpenAI that same day and filed a report with the National Center for Missing & Exploited Children (NCMEC). The account remained active until mid-December – roughly 60 days later.
Following that initial report, our team identified a broader CSAM problem on Sora: dozens of videos showing exploitative content featuring potentially underage individuals. In several cases, the videos appeared to use the likenesses of real minors rather than AI-generated characters, either because the individuals themselves had accounts on the platform or because someone else created accounts using their likenesses from publicly available social media profiles.
As of publication, some of these accounts remain active on the soon-to-be shuttered platform.
“From hate speech to sexually exploitative videos, Sora was quietly a content moderation nightmare,” said Copyleaks CEO Alon Yamin. “With OpenAI increasingly focusing on enterprise tools, this shutdown isn’t entirely a surprise. As it disappears, however, harmful deepfakes and manipulated media will just migrate to platforms that are even more opaque and difficult to audit.”
AI misuse is growing. As it does, independent technology capable of identifying AI-generated images and videos at scale is no longer optional infrastructure. This technology and governance of AI is essential for platforms and companies committed to stopping manipulated media before it causes irreparable harm.
