THE CRUNCH
OpenAI has announced a classifier it says can better detect child sexual abuse material, including images generated by AI, and is offering free access to vetted organisations alongside developer guidance and law enforcement training. Announced at a Monday press briefing in Washington, DC, the tool arrives as AI-generated abuse imagery strains reporting systems: the NCMEC CyberTipline received 400,000 generative AI-related reports in 2025, up exponentially from 67,000 in 2024.
OpenAI says the classifier flags 48 percent fewer images for human review than before, while the images it does identify are twice as likely to contain abusive material. It is designed to catch abusive AI-generated images from a range of tools, including open-source models and so-called nudifying apps, and is described as the first broadly available CSAM classifier with a specific flag for AI-generated material. "It's important for industry to be able to understand what images may contain a fictional child versus a real child," said John Buckley, OpenAI's head of child safety public policy. "That triage is critical in law enforcement reporting flows."
The announcement lands amid mounting political pressure. North Carolina Attorney General Jeff Jackson, cochair of the AG Alliance's AI Task Force, called the influx of AI-generated reports to NCMEC "explosive" and warned that, if it continues, "it is going to mean that we fail in our mission." He described the partnership with a tech company to deter creation of such material as a fairly novel approach, adding: "I would say it's still basically an experiment."
OpenAI is also tightening its own guardrails. ChatGPT's image generator already refuses requests for CSAM, but users who attempt it will now see a warning developed with expert guidance, followed by a 24-hour pause in access and a notice that the account could be permanently closed, alongside support resources. Buckley said the pause reflects expert advice and international evidence on interrupting harmful behaviour. OpenAI says it reports all instances of CSAM to NCMEC.


