Class Action Alleges Grok Trained Deepfake Tools on Real Child Abuse Material

A class action lawsuit filed in the U.S. District Court for the Northern District of California accuses xAI of training the deepfake and "nudify" capabilities of its Grok tool on real images and videos of child sexual abuse. The plaintiffs, listed anonymously, are described as victims tracked through the FBI's Child Exploitation Notification Program.

The complaint alleges that material with known hash values has appeared in deepfakes created with Grok and circulated on X. It further claims that Grok's terms of service treat content posted on X as training material, meaning abuse imagery posted to the site over the past year may have been absorbed by the model.

According to an analysis by the Center for Countering Digital Hate cited in the suit, during an 11-day period between December 2025 and January 2026 Grok created more than 3 million sexualized images, at least 23,000 of which appeared to depict children. The suit contends that xAI's safeguards are weak, diverge from industry practice, and can be circumvented through indirect or euphemistic prompts.

The plaintiffs seek monetary damages and injunctions to prevent further creation of sexualized deepfakes or abuse material. The suit cites "Masha's Law," a 2018 statute that preserves civil remedies for victims of child pornography and online exploitation. A request for comment sent to xAI was not returned.

The case forms part of a wider pattern of criminal investigations and private litigation surrounding the tool, including a separate lawsuit filed earlier this year. For families, the allegations underline how systems that generate and distribute imagery can compound harm to children whose abuse material already circulates online.

This report is based on original reporting by CyberScoop.