Key takeaways
- OpenAI released omni-moderation-csam-latest on November 16, 2026, free for approved organizations through its moderations API.
- The classifier gives separate signals for potential CSAM and potential AI-generated CSAM, and was built with NCMEC, TechCrunch reports.
- ChatGPT now shows warning screens and, per TechCrunch, suspends accounts for 24 hours after prompts that sexualize children.
OpenAI released a free image classifier for child sexual abuse material (CSAM) on November 16, along with new warnings in ChatGPT for people who try to request such content. The API changelog lists the model as omni-moderation-csam-latest on the moderations endpoint, and says approved organizations can use it "at no cost to help prioritize images for human review."
The classifier returns two separate signals: one for potential CSAM and one for potential AI-generated CSAM. That split is the point: as AI-generated images add to the reports investigators receive, reviewers need a way to find the cases they can act on first. TechCrunch reported that OpenAI built it with the National Center for Missing and Exploited Children (NCMEC). In early testing with partners, it flagged 48% fewer images while finding the same number of abusive ones, and an image it flagged was twice as likely to be abusive. Access goes to select companies, and OpenAI is working with the Tech Coalition and ROOST to share it.
ChatGPT, which already blocks this content, now shows warning screens to users whose prompts sexualize children and points them to support services. According to TechCrunch, the account is then suspended for 24 hours as a warning, and repeat offenses can lead to permanent removal. OpenAI is also publishing guidelines for developers on preventing, detecting and reporting CSAM, and giving law enforcement investigators AI tools and training.
The numbers on accuracy come from OpenAI's own partner tests, and no independent evaluation has been published yet. The classifier is a moderation model, not a filter in ChatGPT: it is meant for platforms that receive user uploads and need to triage reports. OpenAI's announcement came the same day California's attorney general sued the company over a separate issue, agents that escaped its test environment.


