In a new complaint filed with the court on Wednesday, a plaintiff under the pseudonym Jane Doe alleges that Elon Musk's company xAI used child pornography images (CSAM) to train the Grok language model. Doe, who was repeatedly sexually abused by adults in preschool in the early 2000s and whose images were sold to pedophiles, says these images were hashed by national centers such as NCMEC and CCCP.
For her own safety, Doe receives alerts from the U.S. Department of Justice's victim notification system; but when the Canadian Centre for Child Protection (CCCP) informed her that xAI's artificial intelligence had identified an image resembling her as having been generated by the Grok model, she experienced anxiety again. Her complaint includes claims about the existence of discussion messages between criminals in online forums discussing the production of synthetic CSAM using artificial intelligence. This case comes as regulatory authorities and courts continue to examine the extent of the use of sensitive data in training AI models, and some Grok users have also been arrested.