xAI has sued one of its own Grok users, Terry Harwood, saying he used the chatbot’s image tools to create non-consensual sexual images and child sexual abuse material, then triggered a contractual duty to protect the company from the legal consequences.
The lawsuit, first reported by Reuters, lands while Elon Musk’s AI company is itself facing claims over Grok-generated abuse images. xAI owns X after Musk folded the social network into the AI company.
In a complaint filed in the Wichita Falls Division of the Northern District of Texas and assigned to U.S. District Judge Reed O’Connor, xAI says Harwood violated the company’s terms of service and acceptable-use rules by using Grok to make sexualized images of people without consent and to generate CSAM. The complaint says Harwood “circumvented safeguards” and distributed abusive material.
xAI’s legal theory leans on an indemnification clause in its terms. The company says users agree to defend and reimburse xAI for claims, losses, costs and legal fees tied to their use of the service, their outputs, violations of the terms or third-party claims stemming from content generated with Grok. xAI asks the court to make Harwood responsible for legal expenses and liabilities arising from his alleged conduct.
The policy case xAI is making
In the filing, xAI presents itself as enforcing a clear boundary around Grok use. The company says its rules prohibit using Grok to violate privacy or publicity rights, undress or “nudify” real people, depict people pornographically, or sexualize and exploit children.
xAI also says it has suspended 52,222 accounts in 2026 and made 73,604 reports to the National Center for Missing & Exploited Children, leading to at least 244 arrests. Those figures are xAI’s claims in the complaint.
The mechanics are not exotic. According to the allegations in related litigation and xAI’s own complaint, users uploaded or supplied images of real people and prompted Grok to alter those images into sexualized outputs. That is the kind of synthetic image abuse platforms have been warned about for years, with the AI model doing the scaling work that used to require more time and skill.
xAI is also a defendant
The company’s move against Harwood comes as alleged victims have sued xAI over Grok-generated non-consensual imagery. CyberScoop reported on a lawsuit alleging that a Wyoming woman, identified as Jane Doe 4, said her stepfather used a childhood photo of her to create more than 7,000 CSAM-related images with Grok and shared them on social platforms.
That lawsuit alleges the stepfather chose Grok because it was less restrictive than other AI systems and would respond to prompts involving a prepubescent minor. It also claims xAI sent a tip to NCMEC but provided only the original photo, and did not provide law enforcement with the generated images or IP address information when requested. Those are allegations, not court findings.
The complaint against Harwood also follows public criticism of Grok’s willingness to generate sexualized edits of real people. Around the turn of the year, Grok users circulated examples of the system producing images that put people, including children, into bikinis or otherwise objectifying contexts. Musk publicly joined one such exchange involving the Ben Affleck smoking meme, asking Grok to replace Affleck’s head with his own in a bikini image. Grok complied, and Musk replied, “perfect,” according to screenshots reported at the time. The image was later removed.
Musk later said people using Grok for CSAM would “suffer consequences.” xAI’s Harwood suit is now the company’s most direct civil attempt to put those consequences on a user, while the separate lawsuits test how much responsibility xAI may bear for shipping a tool that allegedly produced the material in the first place.
This story draws on original reporting from Techdirt.