OpenAI Launches CSAM Classifier and ChatGPT Warnings

Original: OpenAI releases a new image classifier and ChatGPT warnings to combat CSAM

Why This Matters

AI-generated CSAM is an expanding threat; industry-wide classifier sharing sets a precedent.

OpenAI released a CSAM image classifier, ChatGPT deterrent warnings, and law enforcement tools on Nov. 16, 2026, developed with NCMEC, flagging 48% fewer images while catching the same volume of abuse material.

OpenAI announced a package of child safety tools aimed at detecting and deterring child sexual abuse material. The centerpiece is an image classifier built with the National Center for Missing and Exploited Children (NCMEC), distributed free via API to select partners through Tech Coalition and ROOST. Early testing showed it flags 48% fewer total images while identifying the same number of abusive ones — and images it does flag are twice as likely to be genuinely abusive, cutting reviewer workload significantly. The classifier covers both real and AI-generated imagery.

On the product side, ChatGPT will now display warning screens when users attempt prompts that sexualize minors, directing them to support resources and issuing a 24-hour account suspension. Repeat violations can result in permanent removal. OpenAI says it shaped these interventions alongside Protect Children, the Lucy Faithfull Foundation, and University of Tasmania researcher Joel Scanlan, who independently evaluated the deterrent messages.

OpenAI also published new developer guidelines on preventing, detecting, and reporting CSAM, saying early testing prompted several developers to adopt child safety tooling and register with NCMEC. Law enforcement gets dedicated AI plugins, investigative tools, data analysis features, training, and a field guide. The moves come as OpenAI faces ongoing lawsuits over harm to minors and after Common Sense Media rated its technology as 'unacceptable risk' for teens.

Source

techcrunch.com — Read original →