xAI trained Grok on child sex abuse material, a new class action lawsuit alleges

On Aug. 26, 2026, a woman identified in court records as Jane Doe One filed a class action lawsuit against xAI Corp. and xAI LLC, the companies behind the Grok AI model, in the U.S. District Court for the Northern District of California.

The complaint alleges xAI trained Grok on known child sexual abuse material depicting the plaintiff then operated a product that generated and publicly posted new illegal images of children.

Who filed the lawsuit?

The plaintiff is a survivor of child sexual abuse whom the National Center for Missing and Exploited Children identified in the early 2000s. The center assigned her case a series name and added the files to its hash list, a catalog of digital fingerprints investigators use to flag known material. Court papers say she receives federal alerts whenever those images resurface in new investigations.

What the complaint says about Grok's design

Competitors, including OpenAI, Anthropic and Meta, largely block sexual requests while xAI marketed Grok as a system that would answer spicy questions that most other AI systems reject, the lawsuit alleges. Grok's published system prompts reportedly tell the model that "teenage" or "girl" does not necessarily imply underage.

Researchers at the Center for Countering Digital Hate estimated Grok produced more than 3 million sexualized images between Dec. 29, 2025, and Jan. 8, 2026, more than 23,000 of which likely depicted children.

Because Grok lives inside X, it creates and publishes in one step, skipping a checkpoint other platforms have, the proposed class action claims. xAI also treats Grok's outputs as training-eligible by default, allegedly letting illegal images feed the next model, according to the lawsuit.

The allegations involving the plaintiff

Material depicting the plaintiff, traceable through its known hash values, lived in xAI's training dataset, and Grok then generated new images of her, the complaint claims. The Canadian Centre for Child Protection later found AI-generated material depicting her on xAI platforms, per the suit.

xAI reportedly restricted image and video generation to paying subscribers at $40 per month for X Premium+ or $30 per month for SuperGrok. That paywall did not prevent illegal images and, in contrast, ensured the company would profit from them, the lawsuit claims. Grok itself acknowledged "safeguard lapses" that allowed "illegal image gen" on Jan. 11, 2026, the class action contends.

The legal claims

The lawsuit brings two claims:

  • Masha's Law, 18 U.S.C. Sections 2255 and 2252A, a federal civil remedy letting survivors sue over the knowing production, distribution or possession of child pornography
  • 18 U.S.C. Sections 2252A(f) and 1466A, which cover obscene visual depictions of minors

Doe seeks actual damages, liquidated damages of $150,000, punitive damages, and attorneys' fees and costs plus an order requiring xAI to stop generating the material and destroy what it has.

What the case means for people whose images Grok altered

The proposed class covers everyone in the United States whose images or videos taken as minors Grok altered to produce child sex abuse material. xAI has not responded and admitted no wrongdoing.

As of this writing, there is no settlement, no claims process and no money available.