Skip to content

Strengthening the fight against online child sexual abuse with new tools and support

OpenAI

The model detects novel and synthetic child sexual abuse material and provides a signal indicating whether flagged content is AI-generated.

Key points

  • Vetted organizations can access the image classifier through the OpenAI API at no cost.
  • OpenAI estimates the model flags 48% fewer images than prior protections while catching roughly the same volume of abuse.
  • ChatGPT now warns users seeking abusive content, removes the conversation, and pauses account access for 24 hours.
  • Child-safety law enforcement units get dedicated AI tools to parse chat logs, identify connections, and construct timelines.

Why it matters

Child safety teams and developers gain automated detection for previously unseen and AI-generated CSAM that traditional hash-matching systems miss.

Sources