ChatGPT locks out abuse seekers for 24 hours
OpenAI released a tool on the 16th of November 2026 that finds child sexual abuse images, including ones made by AI. Vetted groups can use it for free through the OpenAI API.

OpenAI child safety image toolOpenAI released a tool on the 16th of November 2026 that finds child sexual abuse images, including ones made by AI. Vetted groups can use it for free through the OpenAI API.
OpenAI said it is the first widely available tool of its kind that also flags whether such images may be AI made.
How it works
The tool, called a classifier, looks for visual signs of abuse in images. It gives each image a score so human reviewers know which to check first.
OpenAI built it with help from the National Center for Missing and Exploited Children. That US nonprofit runs a national tip line for child abuse.
New warnings in ChatGPT
ChatGPT will now warn people who try to find abusive material and point them to help. It will remove the chat and pause the account for 24 hours, OpenAI said.
Repeat cases can lead to longer bans, account closure or reports to police.
Below we break down the test results
In OpenAI's test, the new tool flagged 48 percent fewer images than its old tool. It still found about the same number of abusive images.
That means each flagged image was about twice as likely to show abuse, OpenAI said. For SafetyKit, a partner, it caught 90 percent of cases in drawings and cartoons.


What is confirmed and what is not
These numbers come from OpenAI and its partners. Outside groups have not published their own tests yet.
Key facts
OpenAI released the tool on the 16th of November 2026.
Vetted groups get it free through the OpenAI API.
ChatGPT pauses offending accounts for 24 hours.
The tool flagged 48 percent fewer images in tests.
Source: OpenAI: Strengthening the fight against online child sexual abuse with new tools and support