Jane Doe Sues Elon Musk’s xAI Over Alleged Use of Child Pornography in AI Training

A woman known as “Jane Doe” has filed a lawsuit against Elon Musk’s xAI, claiming the company trained its Grok AI chatbot using child pornography that depicts her. This case appears to be the first of its kind accusing xAI of utilizing child sexual abuse material (CSAM) for training purposes.

The lawsuit, reported by Ars Technica, focuses on abuse Doe endured as a preschooler in the early 2000s, during which adult men exploited her to create images later sold to pedophiles online. The National Center for Missing and Exploited Children (NCMEC) and the Canadian Centre for Child Protection have hashed those images to track known child pornography so it can be identified and eliminated when it appears.

Doe receives notifications from the U.S. Department of Justice Victim Notification System whenever her abuse material surfaces. Recently, the Canadian Centre for Child Protection informed her that AI-generated CSAM, resembling her, had been identified on xAI. The lawsuit notes that offenders on online forums were discussing the creation of AI-generated CSAM of Doe and other known victims of CSAM.

The legal action alleges that xAI stores and reuses images generated by Grok to enhance its training model. Doe’s lawyers describe this material as directly related to the CSAM depicting her, suggesting it has been incorporated into Grok’s outputs. The complaint quotes that “CSAM depicting Plaintiff with its longstanding well-known hash values has been used as a part of the dataset used by xAI.”

Additionally, findings from a previous Kaiser report highlighted concerns about AI training datasets containing child pornography. A study by the Stanford Internet Observatory, in association with anti-abuse organizations, discovered over 3,200 images of suspected child sexual abuse within an AI database called LAION. This database has played a crucial role in training prominent AI image creators like Stable Diffusion.

This troubling discovery has raised significant concerns among various sectors, including educational institutions and law enforcement. Child pornography found in these datasets has allowed AI systems to produce explicit and hyper-realistic images of fictitious children, as well as transform social media photos of real teenagers into deepfake nudes. Initially, it was believed that such imagery was generated by mixing adult pornography with innocent images of children. However, the direct incorporation of explicit child images presents a much more alarming reality.

Much of Doe’s legal argument hinges on how Grok’s terms of service pertain to user-generated content. The complaint argues that Grok automatically categorizes public posts on X, along with its own outputs, as training data. It states, “Because Grok’s terms treat public X posts and Grok’s own outputs as training data by default, publicly posting an image does not just expose it to viewers, but also feeds [it] directly into the pipeline xAI uses to train and improve its model.” While xAI does filter out violent content, its terms do not explicitly exclude CSAM or other inappropriate or non-consensual material.

The lawsuit raises a complex question: how can one reverse the damage caused by incorporating abusive material into a trained model? The complaint highlights that “because full removal of a training example’s influence from an already-trained model is technically difficult and not something that xAI has publicly claimed to have done, any CSAM ingested into training before takedown likely continued to shape the model’s outputs even after the original images were removed from public view.”

Doe’s case accuses xAI of violating federal child pornography laws and Masha’s Law, which allows CSAM survivors to seek legal action regarding the production, possession, and distribution of abusive material. One of Doe’s attorneys stated that “xAI did all three” in a press release.

The lawsuit seeks financial compensation for every victim who can demonstrate that Grok generated CSAM based on their actual photographs. Additionally, it requests the court to mandate xAI to dispose of all Grok-generated CSAM currently in storage or utilized in training, and to prevent xAI from generating further CSAM in the future. Achieving this, the complaint suggests, could entail blocking all sexualized outputs from Grok, including non-consensual intimate images and the NSFW “bikini pics” promoted by Elon Musk.

Doe’s attorney expressed that Doe has lived for nearly twenty years with the knowledge that images of her traumatic experience are circulating among predators online, constantly at risk of resurfacing. The attorney emphasizes that xAI must be accountable for knowingly training its models using images of the horrifying abuse she and many other survivors have endured.