OpenAI has released a new image classifier designed to detect child sexual abuse material, including content generated by artificial intelligence. The tool is built in partnership with the National Center for Missing and Exploited Children and will be available to select companies through the OpenAI API. The company is also providing law enforcement with dedicated plug-ins and training for child safety investigations.

The classifier operates with a lower false positive rate than previous models, flagging 48% fewer images while identifying the same number of abusive images. This improvement aims to reduce the burden on content moderators and human reviewers who must manually check flagged content. OpenAI is integrating the tool into its safety systems to prevent the generation and distribution of harmful material.

Users attempting to access CSAM on ChatGPT will now encounter a warning screen and have their account suspended for 24 hours. This update represents a significant escalation in the company’s enforcement measures against severe policy violations. For developers and safety professionals, the availability of this tool through the API offers a more efficient way to filter harmful content at scale.