Generative artificial intelligence (AI) is now capable of creating realistic-looking images and videos with just a text prompt. This capacity has been misused by child sexual abuse and exploitation (CSEA) offenders to generate child sexual abuse material (CSAM), either by altering a real child’s image to “morph” it into sexually explicit content or by generating sexually explicit content depicting imaginary children who do not actually exist.
Federal law has long prohibited “morphed” CSAM as well as so-called child obscenity. However, a 2002 Supreme Court decision ruling computer-generated CSAM is protected speech if it does not involve a real child has caused concern that the First Amendment will protect CSAM created with generative AI and leave lawmakers and prosecutors powerless to combat this novel scourge.
We present a dataset of 71 criminal prosecutions in federal district courts around the country, filed between 2022 and the end of 2025, that somehow involve AI-generated sexually explicit imagery of children, real or imaginary. We believe this is the largest public collection to date of AI-generated CSAM and child obscenity cases in federal court. Analyzing these cases, we show that existing federal criminal laws are sufficient to successfully prosecute AI-generated child sexual abuse material and child obscenity. Most of the cases in our sample have ended in a conviction, prison sentences are higher than average for CSAM and obscenity cases, and the First Amendment is not an obstacle to bringing offenders to justice.
Authors: Riana Pfefferkorn,Stanford University - Stanford Institute for Human-Centered Artificial Intelligence
Cyndie Demeocq, University of Edinburgh
Download paper at SSRN.