OpenAI released the OpenAI CSAM Image Classifier, a new API model for approved organizations that need to prioritize image review for potential child sexual abuse material, including AI-generated material. The company says the classifier returns separate risk signals to help trained reviewers sort cases faster, while still requiring human judgment and existing reporting procedures. That matters because it gives trust-and-safety teams a new triage layer for a problem that is getting harder as synthetic imagery improves.

The classifier is available through the OpenAI API at no cost, but only after application, verification, onboarding, and terms acceptance. The practical takeaway for developers and safety teams is that this is a specialized moderation workflow, not a drop-in consumer feature; access is limited and operational use still depends on human review.