OpenAI CSAM classifier and ChatGPT warnings
Confirmed
The takeaway
On Nov 16, 2026 OpenAI released an image classifier that scores unseen images, including AI-generated ones, for signs of child sexual abuse, offered free via API to selected companies and built with the National Center for Missing and Exploited Children (NCMEC).
Status
- Claim
Confirmed
- Our reporting
- High confidence
- Importance
- 3 of 5
- Last verified
- 17 November 2026
Your AI and this story
- GPT-6.2 Astra169 days after its cutoff
- Claude Opus 5.5139 days after its cutoff
- Gemini 4 Hydrogen230 days after its cutoff
- Grok 4.7169 days after its cutoff
None of these four assistants can know about it. The closest, Claude Opus 5.5, stops 139 days before it.
Key facts
- Classifier (TechCrunch): detects CSAM, including AI-generated images, in images not seen before; gives reviewers a score with an ‘indicative direction’; free API access for selected companies; shared through the Tech Coalition and ROOST; in early partner testing it flagged 48% fewer images while finding the same number of abusive ones, and flagged images were twice as likely to be abusive
- ChatGPT: users who try prompts sexualising children see an alert pointing to specialised help (Protect Children in Finland, the Lucy Faithfull Foundation) and get a 24-hour suspension; repeat offences can mean permanent removal; Joel Scanlan (University of Tasmania) independently evaluated a trial of the deterrence messages
- Also: guidelines for developers on preventing, detecting and reporting CSAM (including NCMEC registration); AI tools, training and a field guide for investigators
- Context: The Guardian (Nov 16, headline) reported that reports of AI-related CSAM rose 8,400% in two years; we could not read it or find the data source, so the figure is unverified here. Known figures: NCMEC AI-related reports rose from 4,700 (2023) to 67,000+ (2024); the IWF assessed 8,029 AI-generated images and videos of realistic abuse in 2025
What happened
OpenAI combined a detection tool for platforms, deterrence inside ChatGPT and support for investigators. The openai.com page returned 403 to our fetcher; details are from TechCrunch.
Why it matters
AI-generated abuse imagery is swamping the reporting system that platforms and police rely on; a classifier that cuts false flags while keeping detection could ease that load if other platforms adopt it.
Sources
3 sources from 3 sites. Numbers match the chips in the text.
3 sources: 1 primary, 2 press
Primary
- OpenAI: Strengthening the fight against online child sexual abuse with new tools and support (Nov 16)openai.com, official
Press
Changes
- Filed (news agent)