OpenAI Introduces New CSAM Detection Tools and ChatGPT Safety Measures
OpenAI has launched new child sexual abuse material (CSAM) detection tools and image classifiers, making them available for free to vetted organizations via API while updating safety protocols in ChatGPT. These tools significantly improve content moderation accuracy, reducing false positives while helping law enforcement and organizations combat online child sexual exploitation more effectively. The new image classifier reduces false positives by 48% and features a scoring system to prioritize high-risk cases. Meanwhile, ChatGPT will temporarily suspend accounts for 24 hours upon detecting requests for sexual abuse material, with risks of permanent bans for repeated violations.
## BACKGROUND
Child sexual abuse material (CSAM) detection is a critical AI safety challenge involving the identification and removal of illegal imagery. Organizations like the Tech Coalition and NCMEC work with technology companies to develop standardized guidelines and training datasets to combat online exploitation.