Elon Musk's AI company is asking a federal judge in San José to freeze a proposed class action accusing its Grok chatbot of manufacturing non-consensual sexual deepfakes — to pause it while related federal criminal proceedings play out, and to ship the case 1,700 miles east to a Fort Worth courthouse the company has steered its users toward for two years.

The fight before U.S. District Judge P. Casey Pitts is, on its face, about venue and fine print. But it turns on a question with stakes far beyond the four pseudonymous plaintiffs: whether a company can write its terms of service to route lawsuits to a favorable bench, and whether a click of "Got it" signs away a user's right to sue in their own state. For now, the case is staying in the Bay Area — and the accusers are staying anonymous.

The defendant is xAI, the chatbot maker Musk folded into SpaceX in February and rebranded as SpaceXAI in July. According to court filings described by KQED, the four plaintiffs say Grok — xAI's chatbot, wired into the social platform X — took ordinary photos of them and turned them into sexualized fakes that were posted publicly. Musk himself is not named as a defendant.

The people behind the pseudonyms

Three of the plaintiffs are X users who say they were harmed in the first five days of January. Ohio Doe posted a photo on Jan. 1 and says Grok created and publicly posted at least six altered images of her within two days. New Jersey Doe posted a shirtless photo on Jan. 4 and found two doctored versions the next day. A South Carolina woman said she uploaded a fully clothed photo on Jan. 2 and discovered the following day that Grok had generated and posted an image of her. Each reported the images to X.

The fourth plaintiff, identified in filings as South Carolina Roe, is not alleged to be an X user at all. She was told on Feb. 26 that her father had been arrested on charges tied to possession and distribution of child sexual abuse material, according to xAI's own motion; the court records say he used Grok to alter images of her as a minor to make it appear they "had just consummated a sexual act." Those images were not shared on X.

The alleged harms trace a product timeline. Grok's image editing was folded into X in December 2024 for premium subscribers, who could tag @grok and ask it to alter a photo, with the result posted publicly in reply. Users, plaintiffs say, quickly began issuing prompts like "Remove her clothes." When image generation opened to all X users in December 2025, plaintiffs' lawyers allege, Grok began churning out hundreds of thousands of increasingly explicit images of women and minors. xAI restricted the feature again on Jan. 8.

A fight over the fine print

xAI's core argument is contractual: the three X users clicked a "Got it" button on an update notice, twice each, and kept using their accounts through June — conduct the company says its terms treat as consent to a clause requiring suits be filed in northern Texas. It notes three judges in the Northern District of California have already enforced that Texas clause.

X has directed users to sue in northern Texas since November 2024 — a district that pointedly does not include Bastrop, where the company is headquartered. Fort Worth's small federal bench includes Judge Reed O'Connor, who has presided over two prior Musk-related suits and whose 2023 financial disclosure showed Tesla stock, according to NPR. X did not answer requests for comment on why it chose those courts.

"Basically they're doing a form of judge shopping," Eric Goldman, a law professor at Santa Clara University, told KQED. "They think that the judges in that particular district are more likely to go in their favor."

Plaintiffs' attorneys from the firm Berger Montague counter that xAI never proved the plaintiffs agreed to the terms — producing no screenshot of either sign-up screen and never specifying which plaintiff saw which notice. They note New Jersey Doe opened his account in March 2009, when the terms then in effect contained no clause binding users to future revisions. Goldman said whether users are bound by updated terms remains genuinely unsettled in internet law, with courts "inconsistent on this issue" and much turning on interface design: "The details of how the user interface appears matter a ton."

The judge also rejected xAI's motion to overturn an earlier order allowing the plaintiffs to proceed under pseudonyms. The venue dispute lands as California Attorney General Rob Bonta and European regulators pursue their own investigations into whether Grok's generated images broke the law.