According to a statement published on OpenAI's website, the new image classification tool analyzes visual indicators to identify previously undetected images containing child abuse. The tool provides review teams with scores indicating which images should be assessed as a priority.
The company said it would provide the system free of charge to approved organizations. OpenAI described it as the first widely accessible classification tool to offer a dedicated signal for AI-generated online child abuse content.
24-hour access suspension on ChatGPT

In addition to its existing safety measures, OpenAI will introduce new warning mechanisms on ChatGPT to stop attempts to search for child abuse content. Users will be told that such content causes harm even when it has been generated by artificial intelligence.
The flagged conversation will be removed, and the user's access to ChatGPT will be suspended for 24 hours; during this period, expert support will also be offered to help end harmful behavior. If violations are repeated, longer restrictions, account closure or notification of the authorities may follow.
The company said it developed this approach with contributions from organizations working to prevent child abuse and academic advisers.