Skip to content
NewA calmer, faster ZeroHour

OpenAI releases CSAM classifier and ChatGPT warning blocks

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

TechCrunch · AIby Ivan Mehta ()2 min read
Part of a story covered by 3 sourcesOpenAI offers free CSAM image classifier and ChatGPT account pausesMerged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Select companies get the classifier free through an API, and blocked ChatGPT prompts can trigger a 24-hour suspension.

OpenAI announced tools meant to detect and deter child sexual abuse material, including AI-generated images. A new image classifier, developed with NCMEC, scores unseen images for visual signs of abuse and is being shared at no cost with select companies through an API, alongside Tech Coalition and ROOST. ChatGPT will warn users who attempt prompts that sexualize children, point them to support, and suspend the account for 24 hours, with permanent removal possible for repeats. OpenAI is also publishing developer guidelines and offering law-enforcement tools and training.

Unconfirmed: The classifier’s accuracy figures come from OpenAI’s early partner testing, not an independent public benchmark.

Key facts
  • OpenAI says tests flagged 48% fewer images while finding the same abusive ones.
  • Images flagged by the classifier were twice as likely to be abusive, OpenAI says.
  • ChatGPT already blocks CSAM content and related requests.
  • Developer guidelines are meant to help report CSAM to NCMEC.
  • Law-enforcement access is managed by program administrators.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

2 min read at techcrunch.com

ChatGPT will ban an account trying to access CSAM content for 24 hours as a warning

TechCrunch · AI

By Ivan Mehta · techcrunch.com

Read the full article at TechCrunch · AI

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.