we_are_coded.by CODE · The world, decoded
БГ

Concept

CSAM detection

OpenAI released a classifier that detects potential CSAM, including AI-generated material, with the first separate signal for whether the material may have been made by a machine. It is free to vetted organizations.

Updated on 17 November 2026 · Checked on 17 November 2026

Until now, detecting child sexual abuse material (CSAM) rested on one thing: a match with a file already known. A generated image matches nothing, because it is new every time. OpenAI released a classifier that looks at the visual signs of abuse itself, instead of hunting for a known copy.

The classifier returns scores that help reviewers decide which images to look at first. It is offered free to vetted organizations through the OpenAI API. According to OpenAI, this is the first CSAM classifier with broad availability that carries a separate signal for whether the material may be AI generated.

The numbers come from OpenAI and its partners, not from an independent test. According to OpenAI, the new classifier would flag 48% fewer images while finding roughly the same number of abusive images, and the images it flagged were about twice as likely to contain abuse. According to fal, reviewers confirmed 72% of high-confidence flags as novel CSAM. According to invideo AI, one in four flagged cases led to a ban. According to SafetyKit, the classifier detected 90% of cases, with false positives on legal adult content below 1%.

OpenAI's teams trained and evaluated it with help from NCMEC, the nonprofit that runs CyberTipline. Access is requested through a form, and the page does not say which organizations get it.

Primary sources

OpenAI's announcement

The visual is generated code-art, no third-party images.