AI-Generated Child Sexual Abuse Images Are Rising: What the Grok Lawsuits Mean for Prevention

An ordinary childhood photo of a young girl, the kind that sits in every family's album.

(Tilde Oyster/The Washington Post; iStock)

According to a federal lawsuit described by The Washington Post, a woman in Wyoming learned that her stepfather had uploaded a harmless childhood photo of her to Grok, the AI chatbot from Elon Musk's company xAI. Within weeks, he had created and traded online more than 7,000 realistic child sexual abuse images — with her face. Now an adult, she joined a case against xAI in July, as Jane Doe 4.

The original suit was filed in March in a California federal court by three teenagers from Tennessee, one of them still a minor. It's seeking class-action status on behalf of everyone in the U.S. whose childhood images were turned into sexualized content using these AI tools.

Jane Doe 4 was never touched. The bodies in those images are not her body. And yet the harm is real — the sleeplessness, the hypervigilance, the specific dread of not knowing who has seen them or where they are now. Survivors of image-based abuse describe symptoms that look like what we see after contact abuse, and the permanence and unknown number of abusers accessing the material exacerbate those harms.

This is not a rare case

The Center for Countering Digital Hate looked at images Grok generated over 11 days in late December and early January, after Musk promoted a feature letting X users edit any posted image with one click. From a sample of 200,000, they estimated roughly 3 million sexualized images in that window, about 23,000 of which appeared to depict children. That's a sexualized image of a child produced roughly every 41 seconds, for eleven straight days, from one product.

X limited the feature to paying users on January 9 and added restrictions on undressing images of people on January 14. After hundreds of thousands of images were created. Since then, the EU has opened a Digital Services Act investigation into X over the feature, California Attorney General Rob Bonta opened a state investigation into xAI, and at least ten countries have taken some form of regulatory action.

What prevention needs to do now

Update the curricula we already teach. If your child sexual abuse prevention training was written before 2025, it says nothing about AI image generation, and your facilitators are walking into schools without language for what kids are actually dealing with. They need to know about the risks and how to get help if it happens to them. Of course, never allowing even family members to take or access photos of you is not a realistic defense against this threat. There is virtually nothing you, as an individual, can do to protect yourself fully. The lessons here need to be about how to get help and, relatedly, the risks of sextortion that can accompany these threats.

Aim prevention at people who might cause harm, not only people who might be harmed. We are still mostly teaching kids not to send sexual photos to others, even people they think they know. Almost no one is running messaging that says plainly: making these images of a real person is abuse, it is a crime under federal law and in most states, and you are not anonymous. Jane Doe 4's stepfather did not suddenly develop an interest in child sexual abuse material. Grok removed the friction, the skill requirement, and the perceived risk that stopped him from creating it sooner. We need to put those barriers back.

Treat product design as a prevention setting. A default setting reaches more children in an afternoon than any of us reach in a year. The eleven days that produced 23,000 images were a design decision, and so were the restrictions that followed five weeks later. Our field should be in the room for the first decision, not commenting on the second. But Musk is not going to invite us inside anytime soon, so we will need to be outside the room but loud.

Use the policy that now exists. The TAKE IT DOWN Act's platform requirements took effect May 19, 2026. Any person depicted in a nonconsensual intimate image, including an AI-generated one, can demand removal, and covered platforms have 48 hours to take it down and remove known copies. The FTC can fine them $53,088 per violation. Most parents and most victim advocates do not know this yet. Let's teach it. Put the request process in your intake materials and your training (for parents and advocates, not kids).

Support the litigation, and say why out loud. The xAI suits matter beyond the five plaintiffs. Regulators move slowly, and fines are survivable, but liability changes what a company's lawyers will let it ship. When building a product without guardrails becomes more expensive than building one with them, the guardrails appear before launch instead of five weeks later, once the harm is done. That is prevention at the source, and our field should be filing briefs, offering expertise, and backing the state AG investigations rather than watching from the sidelines.

How to protect your family

Tell your kids now, before anything happens, that if a fake image of them ever surfaces, they are not in trouble and you will handle it. If it does happen: save what you see, don't delete it, and take a screenshot of where it's posted. Report to the platform under the TAKE IT DOWN process and to the CyberTipline at report.cybertip.org. Then call your local child advocacy center. They know the next step, and you do not have to figure this out alone.