An expanded proposed class action accuses Elon Musk’s X and xAI of more than releasing an image model that can be pushed into making child sexual abuse material. The amended complaint says xAI also failed to give investigators the basic account data they needed after Grok flagged one abusive prompt.
The most severe new allegation concerns a girl identified as Jane Doe 4. According to the complaint, her stepfather used Grok to create about 7,000 sexually explicit images and videos from a single photo taken when she was 11. The filing says Grok generated images involving incest and sexual assault before xAI’s child-safety system intervened only after a prompt describing a gang rape.
That prompt produced a report to the National Center for Missing and Exploited Children, which then alerted police, the complaint says. But the plaintiffs allege xAI’s report contained the original non-explicit photo, omitted the generated abuse images and did not include the user’s IP address. The lawsuit says xAI ignored repeated requests for identifying information for weeks.
Police eventually identified and arrested the stepfather after obtaining a warrant and reviewing his devices, according to the filing. The complaint says investigators found the Grok-generated files and alleges he had traded some material online with other predators. Two days after he was released on bail, he died by suicide, according to the lawsuit.
The plaintiffs say the girl has since experienced anxiety, depression and suicidal thoughts. Her lawyers at Lieff Cabraser Heimann & Bernstein and Baehr-Jones Law argue the case reflects a broader failure by xAI to file useful child-safety reports.
Lawyers say reports lacked usable data
The legal teams said NCMEC found in early 2026 that 90 percent of xAI’s CyberTipline reports were not actionable because xAI did not include information that could help law enforcement locate users. NCMEC did not respond to Ars Technica’s request for comment, according to the report.
The amended complaint also adds another minor, Jane Doe 5, who allegedly was targeted by an adult family friend. Her family learned about the images after police arrested the man and charged him in connection with non-AI child sexual abuse material, the complaint says.
Musk has denied that Grok has been used to generate child sexual abuse images. X and xAI did not respond to Ars Technica’s request for comment.
Stability AI added as defendant
The amended lawsuit also names Stability AI. The plaintiffs allege its open-weight image models have powered “nudify” apps and that some perpetrators may have used Grok, tools built on Stability AI models, or both. The complaint cites a June research paper finding that the Stable Diffusion model family drove 42.7 percent of image-based nudification online.
Stability AI did not respond to Ars Technica’s request for comment.
The proposed classes would cover people in the United States whose real childhood images were altered with Grok or with an app built on a Stability AI model to create sexually explicit images or videos in which they remained identifiable. The lawyers estimate that thousands of minors could fall within those classes.
NCMEC warned in March that reports involving generative AI and child sexual exploitation rose sharply in 2025. The organization said more than 1.5 million CyberTipline reports last year indicated some connection to generative AI, while more than 133,000 reports lacked enough information to determine how the technology was used.
This story draws on original reporting from Ars Technica.
