Lawsuit Says xAI Trained AI on Child Abuse Images
A survivor says Elon Musk's AI tool Grok made new child abuse images from her old photos.
A lawsuit filed Wednesday accuses Elon Musk’s AI company xAI of training its Grok models using child sexual abuse materials (CSAM). The plaintiff, identified only as Jane Doe, says her childhood images—originally created and distributed by abusers—were used without her consent to build Grok’s image and video generation tools. Doe’s lawyers say her images, tracked for decades by organizations like NCMEC, were found in datasets used to train Grok, and that Grok can now generate new CSAM based on her likeness.
The complaint also claims xAI’s terms allow Grok to train on public posts and its own outputs, meaning images posted online—even by abusers—could feed back into the system. Doe’s legal team says this creates a harmful loop: old abuse images train Grok, which then generates new abuse images, which are then used to train Grok further. Doe’s lawyers argue that xAI violates federal child pornography laws and Masha’s Law, which lets survivors sue for damages related to CSAM production, possession, or distribution.
Doe is seeking class-action status to represent all victims whose images were used to generate Grok CSAM. She’s asking the court to order xAI to destroy all Grok-generated CSAM it may be storing, block Grok from ever creating CSAM, and pay damages to victims. The lawsuit doesn’t provide deep technical details but focuses on the real-world harm: survivors being re-traumatized by AI-generated versions of their abuse.
xAI has not publicly responded to the allegations. The company’s Grok tool is already under scrutiny for its handling of violent and explicit content, and this lawsuit adds pressure on regulators and courts to investigate how AI companies manage harmful training data.
- A survivor says her childhood abuse images were used to train xAI’s Grok AI, which can now generate new abusive images of her.
- The lawsuit claims xAI violated federal child pornography laws and seeks damages, destruction of AI-generated CSAM, and a ban on Grok creating such content.
- Grok’s terms allegedly allow it to train on public posts and its own outputs, creating a cycle where abuse images fuel more abuse images.
Why It Matters
This case could force AI companies to prove their training data is safe, affecting how AI tools are regulated and protecting survivors from digital re-victimization.