Seven Thousand Images From a Childhood Photo

An amended U.S. class-action complaint filed July 7, 2026, alleges that Elon Musk's Grok chatbot was used to turn ordinary photographs of children into thousands of sexual images. The filing, which adds the artificial-intelligence company Stability AI as a defendant, is the newest escalation in a wave of lawsuits and regulatory investigations spanning three continents over AI-generated sexual imagery.

The claims are among the most disturbing yet to be leveled at a mainstream artificial-intelligence product. In an amended class-action complaint filed on July 7, one plaintiff, identified only as Jane Doe 4, says her stepfather used Grok, the chatbot built into Elon Musk's platform X, to generate roughly 7,000 sexually explicit images and videos of her from a single photograph taken when she was about 11 years old, according to NPR. The generated material, the filing says, depicted her nude and performing sexual acts.

A second new plaintiff, Jane Doe 5, alleges that a man used a photograph from her eighth-grade graduation to produce child sexual abuse material through Grok, NPR reported. The two joined a suit originally brought in March 2026 by three teenagers from Tennessee. The amended complaint also names a new defendant, Stability AI, whose Stable Diffusion image model the plaintiffs say was trained on a dataset containing explicit material and released without adequate safeguards, allowing it to underpin third-party “nudify” tools.

A complaint built on real children's photographs

The lawsuit does not describe images conjured from nothing. It describes ordinary pictures — a child at 11, a teenager at a graduation ceremony — fed into a system that, the plaintiffs contend, was designed without effective barriers against turning them into abuse material. The complaint alleges that xAI, the Musk company that makes Grok and is now identified in the filing as SpaceXAI, built the tool without adequate protections to stop the creation of AI-generated child sexual abuse material from photographs of real minors, NPR reported.

The pattern the suit describes did not begin in a courtroom. Grok's image tools drew international alarm at the start of 2026, when users found the chatbot would “digitally undress” real people on request. CNN reported in January that the feature was being used to strip clothing from images of women without their consent, prompting an outcry, according to CNN Business. The Center for Countering Digital Hate estimated that Grok generated some 3 million sexualized images of women and children over roughly ten days at the end of December and start of January, a figure cited by Al Jazeera.

A reporting system investigators say failed

Central to the amended complaint is the allegation that the company repeatedly failed to properly report the material to authorities. By early 2026, the National Center for Missing & Exploited Children found that about 90 percent of the tips xAI submitted to its CyberTipline were not actionable by law enforcement because the company declined to include the user information investigators would need to identify and locate offenders, NPR reported.

In Jane Doe 4's case, the complaint says, the company sent a single tip to the center in February — only after her stepfather asked the model to generate an image of the girl being raped by multiple men. The suit frames those gaps not as isolated lapses but as a systemic failure to treat automated abuse the way established platforms are expected to treat it: by preserving evidence and passing it, with identifying detail, to investigators.

Regulators on three continents

The litigation is only one front. The European Commission opened formal proceedings against X in late January under the Digital Services Act, examining whether the platform had mitigated “risks related to the dissemination of illegal content in the EU, such as manipulated sexually explicit images, including content that may amount to child sexual abuse material,” CNN reported. The Commission has also examined Grok separately over the generation of what it called child-like deepfakes, according to Al Jazeera.

The pressure has come from several directions at once. A European privacy watchdog opened what CNN described as a large-scale investigation into the images, CNN reported, and a Paris prosecutor's investigation led to a search of X-related premises in early February, according to CNN. In the United Kingdom, the regulator Ofcom opened a formal investigation into whether X was doing enough to keep illegal deepfakes off its service, CBS News reported.

In March, an Amsterdam court went further, ordering xAI to stop generating and distributing sexual imagery of people who had not given explicit consent and warning of fines of 100,000 euros a day for noncompliance, Al Jazeera reported. The company had argued in court that it was impossible to fully prevent abuse; the judge was unpersuaded.

The company's response

Under mounting criticism, the company said in January that it would restrict Grok's image generation and editing to paying subscribers and bar the tool from producing sexual images of real people, Al Jazeera reported. NBC News found in the following weeks that the chatbot was still producing sexual deepfakes despite those pledges, according to NBC News. X has said it removes child sexual abuse material, suspends the accounts responsible and works with law enforcement, and that anyone who uses or prompts Grok to make illegal content faces the same consequences as uploading it, as reported by NBC News. The company did not respond to requests for comment on the amended lawsuit, NPR reported.

What the amended complaint adds to that record is not a new abstraction but new names and new numbers: two more plaintiffs, a second corporate defendant, and a single childhood photograph that, the filing says, became seven thousand. Researchers and child-safety authorities have warned that generative tools collapse the distance between an innocuous image and its weaponization — and that the burden of policing the result is landing on courts and regulators rather than on the systems that produce it.