OpenAI launches new image classifier and ChatGPT warnings to curb CSAM
Reported by AndroGuider, TechCrunch, TechTimes Vietnam and 3 more outlets6 sources ↓
OpenAI announced a free, API-accessible image classifier for selected partners, alongside ChatGPT warnings and 24-hour account suspensions for prompts seeking content that sexualizes children.
Key points
- OpenAI says the classifier flagged 48% fewer images while identifying roughly the same number of abusive images in tests
- Images flagged by the classifier were about twice as likely to contain abuse, according to OpenAI
- Repeated violations may lead to longer restrictions or permanent account closure
- OpenAI also published CSAM-prevention guidance for developers and tools and training for law-enforcement investigators
Sources
- AndroGuiderRead ↗OpenAI Launches CSAM Image Classifier and ChatGPT Warnings to Strengthen Safety
- TechCrunchRead ↗OpenAI releases a new image classifier and ChatGPT warnings to combat CSAM
- TechTimes VietnamRead ↗OpenAI launches new image classifier and ChatGPT warnings to curb CSAM
- Crypto BriefingRead ↗OpenAI rolls out image classifier and ChatGPT warnings to fight CSAM
- SQ MagazineRead ↗OpenAI Launches New CSAM Detection and ChatGPT Warnings
- TechBookyRead ↗OpenAI Releases Child Safety Image Classifier and New ChatGPT Warnings
Details
The classifier is intended to flag suspected abuse in previously unseen images, including AI-generated material, and assign risk scores to help trained reviewers prioritize cases. OpenAI says it developed the tool with the National Center for Missing and Exploited Children and is working with the Tech Coalition and ROOST to extend access. The scores are indicators, not legal findings or proof of an image’s origin.
In tests reported by OpenAI, the classifier flagged 48% fewer images for review while identifying roughly the same number of abusive images as an existing tool; images it flagged were about twice as likely to contain abuse. TechBooky reports these results came from OpenAI’s evaluation of images uploaded by users. The company says organizations using the tool should test thresholds on their own data, and trained people should assess flagged images.
ChatGPT warnings appear when users submit prompts seeking content that sexualizes children. OpenAI says the warning explains the harm and points users toward specialist support; the flagged conversation is removed and access is paused for 24 hours, with longer restrictions or account closure possible for repeated violations. OpenAI also released developer guidance and is providing law-enforcement investigators with tools, training and a field guide.
Written from coverage as of
Images
- TechCrunch
- AndroGuider
- TechTimes Vietnam
- TechBooky