Until now, the lawsuits stacking up against xAI over Grok have all leaned on one theory: that the company launched without the guardrails its rivals had already built. A complaint filed earlier this week goes further, and the allegation is uglier. It accuses xAI of training its models on child sexual abuse material.
No one had leveled that particular accusation at Elon Musk’s company before. It’s not a small difference, either. Leaving out a filter is negligence. Putting CSAM into a training pipeline is a category of its own.
How Grok got here
Pressure has been building on the chatbot for a couple of months, dating back to the moment Musk used his own X account to tout Grok’s photo-nudifying capability. The aftermath amounted to a kind of golden age for online perverts. Millions of nonconsensually created and distributed sexual deepfakes flooded the X timeline. Children accounted for a large share of the nudification victims.
Regulators opened investigations quickly. Individual suits followed, along with no fewer than three class actions. Musk’s answer was to narrow access to the feature, which amounted to sticking it behind a paywall and generating still more revenue from it. The so-called “spicy mode” hasn’t gone anywhere. It simply carries more restrictions now.
Then came a stranger turn. xAI took two of its own users to court over their use of the tool to produce CSAM. In the same filing, the company stated that it had assisted in the arrest of at least 244 individuals who created or distributed CSAM through Grok.
The plaintiff is already in an FBI database
Behind the proposed class action is a plaintiff identified only as Jane Doe. She was previously victimized by child pornography and is tracked through the Federal Bureau of Investigation’s Child Exploitation Notification Program. According to her, CSAM depicting her as a child was fed into Grok’s training, and that training then generated fresh CSAM of her and of other victims, this time synthetically produced.
“xAI must be held responsible for knowingly training its models on images of the horrific abuse she suffered, and on the abuse images of every other survivor in this class,” said Sarah London, the plaintiff’s counsel and a partner at Girard Sharp law firm, in a press release.
The loop the complaint describes
The mechanism spelled out in the suit deserves a second read. According to the complaint, xAI deliberately built Grok to answer user prompts for sexual content as a way to “entice more users of both X and Grok.”
The lawsuit says xAI’s own policy treats anything published publicly on X, plus any Grok output, as fair game for training. The consequence: every piece of CSAM Grok posted fed directly back into the model responsible for it.
A closed loop. Output becomes input becomes output.
“Therefore, as long as Grok retains the capability to generate this kind of content, the harm cannot be considered resolved just because individual images are removed or content is no longer posted publicly,” the complaint states. “If such material entered training, its influence likely will persist and contribute to future abusive outputs, exacerbating the trauma that survivors face as victims can have their abuse or likeness reproduced, transformed, and redistributed at scale, repeatedly stripping them of control over imagery tied to their exploitation.”
What she’s asking for
The first request is damages for victims whose images were allegedly used to produce CSAM. The second is the one that bites: destruction of every piece of Grok-generated CSAM, covering not just what remains publicly visible but anything capable of training the models.
Taking down a post costs almost nothing. Scrubbing a training corpus does, and that sort of remedy pushes a company into demonstrating exactly what its data contains.
“Possessing CSAM is a crime, producing CSAM is a crime, and distributing CSAM is a crime. xAI did all three,” said Margaret E. Mabie of Marsh Law Firm, another of the plaintiff’s attorneys, in the press release. “There is no artificial intelligence exception to federal child protection laws and xAI must face the consequences of its reckless actions.”
That framing from Mabie is the one that should keep xAI up at night. By its own account in court, the company helped land 244 people in handcuffs over material Grok produced. The new complaint wants to know why the system that produced it is measured by another yardstick.