Summary: The question now sitting at the center of the Grok nonconsensual-deepfake class action is no longer what the model generated, but whether the people harmed by it can keep their names out of the case. Which makes the stakes stark — if the motion is granted, the lawsuit could lose the very people bringing it. WIRED reported on June 3, 2026 that xAI has gone to the US District Court for the Northern District of California seeking to overturn the earlier pseudonymity ruling, so that all four plaintiffs would have to litigate under their real names.
The question now sitting at the center of the Grok nonconsensual-deepfake class action is no longer what the model generated, but whether the people harmed by it can keep their names out of the case. WIRED reported on June 3, 2026 that xAI has gone to the US District Court for the Northern District of California seeking to overturn the earlier pseudonymity ruling, so that all four plaintiffs would have to litigate under their real names. The dispute traces back to January, when Grok was used at scale to produce “undressing” images of real people, and with this move the fight has been relocated: away from the model’s output, toward whether those it was used against may sue anonymously at all.
The plaintiffs’ legal team answered without diplomacy. “Having stripped them of their clothes, xAI now seeks to strip Plaintiffs of their pseudonyms,” Sophia Rios of Berger Montague wrote in a filing. In affidavits dated May 29, all four plaintiffs said the same thing: forced to continue under their real names, they would consider dropping out. Which makes the stakes stark — if the motion is granted, the lawsuit could lose the very people bringing it.
How the Dispute Began
The trouble started in January 2026, when Grok was used to churn out nonconsensual sexualized images at scale, drawing condemnation worldwide. On the Center for Countering Digital Hate’s estimate, about 3 million sexualized images were generated in just 11 days, and roughly 23,000 of them may involve minors. The New York Times followed the fallout as it developed, and the European Union then opened a formal investigation into xAI.
The lawsuit took form inside that same chaos. In January, one claimant suing under a pseudonym brought the case to the Northern District of California; once a judge let the claimant proceed as Jane Doe, the plaintiffs came back in early May with four anonymity-protected claimants — two in South Carolina, one in New Jersey, one in Ohio. On the other side, xAI now sits within SpaceX, which has reportedly set aside more than $500 million for the legal fallout.
What xAI Filed, and Why
Court filings reviewed by WIRED show xAI bringing two motions in mid-May, both built on one demand: the plaintiffs must give their names. The reasoning runs on three levels. Civil procedure, the company holds, normally requires parties to sue under real names, with pseudonymity as the exception. The public, it continues, has a right to know who is suing whom, and anonymity eats away at the transparency of a proceeding. And in its reading, the plaintiffs have produced no concrete evidence of further harm.
On the privacy worry, the filing is strikingly blunt: “Factoring out the deepfake image itself — as it will remain under seal — there is nothing inherently stigmatizing” about the mere fact that such an image once existed. WIRED’s request for comment drew no response from xAI or its lawyers.
The Affidavits: Four Accounts, One Fear
The four affidavits describe different circumstances but circle the same dread.
The Doe from South Carolina is the case’s original Jane Doe. She found a deepfake of herself “stripped down to a revealing bikini,” and afterwards feared how employers would judge her, being doxed, and retaliation from Musk supporters. A fellow South Carolinian, the Roe, alleges Grok created explicit images of her when she was a child. Court records state that her father “was facing criminal charges related to the possession and distribution of CSAM,” and that Grok altered photos “to make it appear as if we had just consummated a sexual act.” Her filing reads: “Publicly identifying me would cause me untold harm.”
The Doe from New Jersey is a man who asked publicly that “Grok not create images of me without my consent” — and the next day came across two deepfakes of himself. The Doe from Ohio is likewise cited in the filings, without her individual details laid out.
What Anonymity Protects
The default in US federal litigation is that parties appear under their own names; in sexual-violence and privacy cases, however, courts may grant pseudonyms, on the simple ground that unmasking is itself an injury. “Forcing plaintiffs in privacy suits to sue in their names does so little for judicial transparency,” Danielle Citron of the University of Virginia School of Law told WIRED, calling compelled naming “unacceptable and unjust” because its practical effect is to stop victims from bringing claims at all.
Seen from the plaintiffs’ side, then, the motion is pressure by design. Rios wrote elsewhere that xAI “suggests that the abuse it has perpetuated is no big deal.” If the court rules for xAI, every deepfake victim will face a fork in practice: abandon the claim, or take on the exposure of being named.
The Takeaways for AI Companies and Victims
Three things stand out. Unmasking motions are turning into a routine defensive play in AI litigation — where a generative product’s harms are sexual and private by nature, insisting on named plaintiffs is the cheapest way to make suing expensive. A second track runs alongside the courtroom: the EU’s formal investigation into xAI needs no victim to step forward, and its existence only throws the pseudonymity problem in court into sharper relief. And for product teams, the case reads as a warning: nonconsensual-deepfake capability is a product defect carrying legal exposure measured in the hundreds of millions, and whether pseudonymity survives will set how often that exposure gets triggered.