United States District Court: The dispute arose from Minnesota’s enactment of HF 1606, an anti-nudification law regulating AI tools capable of generating realistic nude or sexually explicit images of identifiable persons. xAI LLC, the developer of Grok Imagine, filed suit against the Minnesota Attorney General, alleging that the law violates the First Amendment and seeking a preliminary injunction to prevent its enforcement. Donovan W. Frank, District Judge, considered xAI’s motion for a preliminary injunction. The Court held that xAI failed to demonstrate irreparable harm, emphasizing that the company waited nearly three months after the statute was signed into law before initiating the action and sought injunctive relief only shortly before the law was scheduled to take effect.
The Court also found that the balance of harms and the public interest weighed strongly in favour of the State, given the Legislature’s objective of combating the significant harms associated with AI nudification technology. Accordingly, the Court denied xAI’s motion for a preliminary injunction.
Background
The dispute concerned Minnesota’s newly enacted law, HF 1606, which prohibits operators of certain websites, applications, software, programs, and services from permitting users to generate realistic nude images or videos depicting identifiable individuals. The law exempts services that require users to possess technical expertise to create such alterations and authorizes enforcement by the Minnesota Attorney General. Available remedies include civil penalties of up to $500,000 per violation, as well as injunctive and other appropriate relief.
Anti-Nudification Technology
The legislative history of HF 1606 sheds light on the harms posed by AI nudification technology that were considered by the Minnesota Legislature. For example, three women testified about the significant harms they experienced after a perpetrator used nudification technology to create realistic pornographic material depicting them. These harms included missed work, fears for their family members, particularly their children, anxiety about leaving their homes, and the emotional burden of wondering who had seen, or might see, convincing and sexually explicit images and videos of them. This perpetrator created non-consensual pornographic images and videos of more than 80 women.
With respect to children, AI nudification technology is increasingly being exploited to generate child sexual abuse material. The Legislature heard evidence of a marked increase in AI-generated child sexual abuse material on dark web forums as well as on the clear web, including commercial websites. Further, nudification technology presents a real and growing threat in schools. AI software has been used to fabricate sexually explicit images of students, which can then be rapidly disseminated through group chats and social media platforms. In response to the proliferation of these harmful images and the serious harms caused by AI nudification technology, the Minnesota House passed the anti-nudification law, at issue in this case, on April 23, 2026.
RAINN, which operates the National Sexual Assault Hotline, informed the Legislature that in a single month in 2023, nudification apps had 24 million users, with over 95 percent of manipulated images found to be nonconsensual, sexually explicit, and depicting women. The Internet Watch Foundation similarly reported that CSAM tips linked to generative AI rose from 6,800 in the first half of 2024 to over 400,000 by June 2025. A further study, examining the first 11 days after a social media platform launched a one-click AI image-editing tool, found that roughly 65 percent of the 3 million images created were photorealistic sexualised images, generated at a rate of 190 images per minute, approximately 23,000 of them depicting children.
xAI and Grok Imagine
The plaintiff, xAI LLC, develops artificial intelligence products, including Grok and Grok Imagine. According to xAI, Grok Imagine allows users to create images and videos by typing or dictating a brief description of the content they wish to generate, with the requested image or video produced within seconds. Grok Imagine can also be used to create sexualized images through its nudification features. xAI asserts that its Terms of Service prohibit users from engaging in “any illegal, harmful, or abusive activities, or any other behaviour that violates its Acceptable Use Policy.”
Following the enactment of HF 1606, xAI filed suit against Keith Ellison, in his official capacity as Attorney General of Minnesota, alleging that the statute violates the First Amendment. xAI further sought a preliminary injunction to prevent enforcement of the law while the litigation proceeds.
xAI states that in 2026 alone, it suspended over 50,000 accounts and made more than 70,000 reports to the National Center for Missing & Exploited Children, resulting in at least 244 arrests. The State, however, pointed to three pending lawsuits alleging that Grok Imagine was used to generate child sexual abuse material, as well as a fund exceeding $500 million that xAI’s parent company, SpaceX, has reportedly set aside for potential litigation losses tied to Grok Imagine’s ‘Spicy’ and ‘Unhinged’ modes.
Analysis, Law, and Decision
A. Delay and Irreparable Harm
The Court observed that HF 1606 was signed into law on May 7, 2026. Nevertheless, xAI waited nearly three months before filing suit and sought injunctive relief only shortly before the statute was scheduled to take effect. The Court noted that xAI offered no specific explanation for the delay, other than the general assertion that laws such as HF 1606 can take time to come to the attention of corporate management and for management to authorize the expenditure of resources necessary to challenge them.
The Court further stated that xAI is a sophisticated and well-resourced litigant capable of acting promptly when its interests are at stake. In the Court’s view, had xAI genuinely feared irreparable harm, it would likely have moved more expeditiously to commence the action and seek injunctive relief. The Court concluded that xAI’s failure to do so strongly undermined its claim of irreparable harm.
B. Commercial injury and irreparable harm
The Court next addressed xAI’s contention that the prospect of civil penalties of up to $500,000 per violation constituted irreparable harm. However, the Court held that any such penalties would be monetary in nature and therefore recoverable, meaning that even substantial financial penalties do not qualify as irreparable harm. The Court also noted that xAI had already disabled its nudification functionality for users in Minnesota, thereby significantly reducing the likelihood that it would incur penalties while the litigation remained pending.
The Court concluded that these alleged forms of commercial injury did not establish a threat of irreparable harm. It acknowledged that xAI had taken steps to modify its AI tools to ensure compliance with HF 1606 and had incurred costs associated with increased moderation efforts. Nevertheless, the Court found that xAI had failed to demonstrate that these costs were irreparable in nature.
The Court also found xAI’s assertions regarding substantial expenditures to be supported only by vague and conclusory statements rather than concrete evidence. As for the claim that increased moderation could result in the loss of Minnesota users and corresponding business opportunities, the Court characterized that harm as entirely speculative. The Court observed that xAI relied primarily on a single declaration containing only generalized and conjectural support for this claimed injury. Moreover, because HF 1606 applies equally to all companies offering AI tools that enable nudification, the Court found no basis to conclude that xAI would suffer a competitive disadvantage relative to other market participants.
C. First Amendment Claims and Irreparable Harm
The Court also addressed xAI’s argument that the alleged infringement of its First Amendment rights constituted irreparable harm. It emphasized that an asserted constitutional violation does not automatically entitle a party to injunctive relief. Rather, a plaintiff seeking a preliminary injunction must make a clear showing that it is likely to suffer irreparable harm in the absence of such relief. In the Court’s view, xAI’s substantial delay in seeking injunctive relief significantly undermined its claim of urgency and weakened its ability to satisfy this requirement. Accordingly, the Court concluded that xAI had failed to establish irreparable harm and held that this deficiency alone was sufficient to warrant denial of the motion for a preliminary injunction.
E. Balance of Harms and Public Interest
The Court observed that, where a preliminary injunction is sought against the government, the balance of harms and the public interest factors merge. It concluded that both considerations weighed strongly in favour of the State. The Court emphasized that the State had enacted HF 1606 through a democratic process and with near-unanimous legislative support to address the undisputed harms associated with AI nudification technology and the resulting proliferation of photorealistic sexualized images.
The Court further noted that the State’s decision to prohibit the provision of AI tools that enable users to create such images reflected a direct response to the unique harms posed by this emerging technology. In the State’s view, efforts to address violations only after harmful images have been created and disseminated were insufficient to prevent the resulting harms. By contrast, the Court found that xAI had failed to present persuasive evidence of irreparable harm, notwithstanding its assertions that its First Amendment interests were implicated. Weighing the State’s interests against any potential harm to xAI, the Court concluded, without hesitation, that the balance of equities tipped decisively in favour of the State.
F. Constitutional Questions and Denial of Preliminary Injunction
The Court acknowledged that the parties had raised substantial constitutional questions under the First Amendment. It noted, however, that the issues presented are complex, particularly considering the novel technology at issue and the risks it poses to the public. According to the Court, these constitutional questions warrant careful and thorough consideration, which they will receive as the case proceeds.
The Court further observed that a determination of xAI’s constitutional claims would likely occur in the near future, given the State’s pending motion to dismiss, which squarely presents those issues for adjudication. The Court also noted that, at a later stage of the proceedings, it may consider whether xAI is entitled to permanent injunctive relief on the merits. For the time being, however, the Court concluded that xAI had failed to demonstrate irreparable harm and that the balance of harms weighed strongly in favour of the State. Accordingly, the Court denied xAI’s motion for a preliminary injunction.
Thus, the Court denied xAI’s motion for preliminary injunction.
[x.AI LLC v Keith Ellison, 2026 SCC OnLine Dis Crt US 4, decided on 04-09-2026]
