Class Action Lawsuit Against Elon Musk's Grok AI Expands Over Child Sexual Abuse Material Generation

A class-action lawsuit against xAI over its Grok tool expanded in July 2026, adding two anonymous plaintiffs — Jane Doe 4 of Wyoming and Jane Doe 5 of Wisconsin — who allege adult perpetrators used Grok to generate CSAM from real childhood photographs. The amended complaint also names Stability AI as a defendant, alleging its open-weight Stable Diffusion tools were used in the abuse, and accuses xAI of obstructing a law enforcement investigation by filing deficient NCMEC CyberTipline reports that omitted generated images and user IP addresses. The five plaintiffs seek $150,000 per violation per defendant, stronger AI safeguards, and class certification. The case is Doe 1 v. X.AI Corp., No. 5:26-cv-02246 (N.D. Cal.), filed March 16, 2026 and amended July 7, 2026.

Lawsuit expansion with plaintiffs

• ▪The five anonymous plaintiffs accuse xAI and Stability AI of producing child sexual abuse material, negligence, defective product design, and creating a public nuisance.
• ▪A class-action lawsuit against xAI over its Grok tool expanded in July 2026 to include two new anonymous plaintiffs, Jane Does 4 and 5.
• ▪The class-action lawsuit, originally filed in March 2026 by three women, was amended on July 7, 2026, to add the new plaintiffs.

Minor victim allegations

• ▪Jane Doe 5 alleges an adult male relative of a classmate used Grok to convert her eighth-grade graduation photograph into illicit sexual material.
• ▪The stepfather of Jane Doe 4 died by suicide two days after being arrested and charged with child exploitation crimes.
• ▪Jane Doe 4 alleges her stepfather used Grok to generate over 7,000 sexually explicit images and videos from a photograph taken when she was 11 years old.

xAI law enforcement response

• ▪xAI failed to include the generated images or the stepfather's IP address in its NCMEC report, despite multiple requests from law enforcement.
• ▪xAI sent only one tip to the National Center for Missing and Exploited Children in February 2026 regarding Jane Doe 4's images.

Stability AI defendant addition

• ▪The lawsuit alleges Stability AI released Stable Diffusion 1.0 as an open-weight model, allowing downstream developers to bypass safety protections.
• ▪The class-action lawsuit was expanded to name Stability AI as a defendant, alleging its image-producing tools were used to create child sexual abuse material.

Stable Diffusion training data

• ▪A 2023 Stanford study found that the dataset used to train Stable Diffusion models was created through unguided web crawling and ingested explicit material.
• ▪Although Stable Diffusion 1.0 included a classifier to block explicit images, its training data allowed downstream developers to exploit the model.

Model guardrail rollback

• ▪The lawsuit alleges Stability AI rolled back stronger safety guardrails introduced in Stable Diffusion 2.0 in response to user complaints.
• ▪A research paper by scholars from the University of Wisconsin-Madison found a more recent Stability AI model generated a higher proportion of explicit content than Stable Diffusion Version 2.

3 sources