Post

AI-generated NCII is a serious form of image-based sexual abuse, while mitigations may pose risks to expression. A new report examines the policies and technical safeguards of 7 major AI companies, identifying critical gaps in protection. cdt.org/insights/tow...

The research finds that company policies often blur the line between consensual sexual expression and abuse, a conflation that risks both under-protecting victim-survivors and over-restricting lawful expression, while technical safeguards lack rigorous testing to demonstrate their effectiveness.

Protections applied at the product level also leave gaps when third-party developers adapt models, particularly open-weight models. cdt.org/insights/tow...