Elon Musk’s xAI used child porn to train Grok models, lawsuit says

A class-action lawsuit has been filed against the artificial intelligence firm xAI, alleging that the company utilized child sexual abuse material to train its generative model, Grok.

The anonymous plaintiff filed the complaint after a child protection organization identified AI-generated imagery on the platform depicting her childhood abuse. According to the lawsuit, the original hashed media, which had been tracked by safety organizations for decades, was allegedly included in the datasets xAI utilized to develop Grok's media generation features. The legal action highlights the profound trauma experienced by survivors when automated systems replicate and distribute such materials.

The complaint also raises major concerns about the system's ongoing training mechanisms and data policy. It alleges that xAI's default settings integrate public social media posts and Grok's own outputs back into its machine learning pipeline, meaning the model continually trains on illicit outputs. Because purging specific data from an established model is highly complex, the plaintiff argues the model remains corrupted by the original data. The lawsuit seeks financial compensation for affected victims and asks the court to compel xAI to purge all stored abuse material, permanently prevent the generation of such content, and restrict the model from producing sexualized imagery.