MINNEAPOLIS – The women Grok targeted do not have to wait for a final court ruling to get legal protection in Minnesota. A federal judge handed them that much on Saturday, refusing to put the state’s nation-first ban on AI-generated intimate image apps on hold while Elon Musk’s xAI continues its legal challenge.
U.S. District Judge Donovan Frank denied xAI’s request for a temporary restraining order, allowing Minnesota’s law to take effect as scheduled on August 1. The ruling does not end the broader legal fight; xAI will still get its day in court on August 19. But for now the law is operative, carrying $500,000 fines per generated nonconsensual intimate image and covering any app or website capable of producing content “so realistic that a reasonable person would believe” the depicted intimate parts belong to a real, identifiable individual.
The judge’s reasoning turned on a detail xAI would rather no one examine too closely. The company filed its lawsuit on July 29, nearly three months after the law was signed and exactly three days before it was set to take effect. Frank’s logic was direct: a company that waits that long before suing cannot credibly argue that harm is too immediate for a court to deliberate. That delay, he wrote, “suggests that harm is not immediate.” Courts evaluating temporary restraining orders apply a four-part test in which the applicant’s likelihood of irreparable harm weighs heavily. xAI lost that test largely before its attorneys finished arguing their case.
Minnesota’s law is not a response to abstract risk. It follows a specific and documented pattern: in early 2026, users of X exploited Grok’s image generation tools to produce and distribute nonconsensual sexualized images of real women and children. The flood of deepfakes triggered investigations and bans in multiple jurisdictions including California and Indonesia, a UK government condemnation of xAI’s subsequent decision to restrict the Grok feature to paid subscribers rather than eliminate it, and a finding by Canada’s privacy commissioner that xAI violated federal privacy law after Grok generated nearly three million sexualized deepfakes in its first ten days, including 23,000 images of children. xAI implemented policy changes in January. NBC News later reported that similar violations continued through April.
xAI disputes that any of this history justifies what it calls an “overbroad, content-based ban on free speech.” In court filings, the company acknowledged it does not oppose Minnesota’s efforts to address nudified imagery specifically, but argued the law extends well beyond that target. The statute’s language covers apps capable of generating images so realistic they could be mistaken for real individuals’ intimate parts, sweeping in applications the company contends have no connection to nonconsensual content. There are “far less restrictive alternatives” the state could have chosen, according to Engadget.
That argument may yet succeed. It is the kind of First Amendment challenge that courts have occasionally found persuasive when statutes define prohibited categories too broadly. Whether Minnesota’s drafters drew the line in the right place is a question for the August 19 preliminary injunction hearing. The distinction between that and Saturday’s TRO ruling matters: a temporary restraining order demands proof of imminent, unrecoverable harm; a preliminary injunction gets full briefing, competing arguments, and a considered assessment of likely success on the merits. xAI lost the first test on its own scheduling. The second remains genuinely open.
In the meantime, every app store operating in Minnesota now faces legal exposure if it hosts applications that generate nonconsensual intimate imagery of real people. The $500,000-per-image fine is not a deterrent in the ordinary sense. It is a business-ending liability for any developer who ignores the law and gets caught. Whether major platforms will scrub their catalogs voluntarily or wait for enforcement action to clarify the edges of the law’s reach is not yet clear, according to TechCrunch.
xAI’s legal exposure on Grok’s image capabilities extends well beyond Minnesota. The company filed a civil lawsuit against a user two weeks ago, accusing a South Carolina man of using Grok to generate child sexual abuse material. That case positions xAI as a victim of misuse. The Minnesota case positions the company as one asking a court to disable the state law designed to prevent that misuse. xAI is arguing both sides of the same problem simultaneously, a legally coherent position but not an intuitive one.
The more enduring question Minnesota’s law forces into the open is what standard courts should apply when an AI platform’s capabilities are genuinely neutral, harm is produced by user behavior, and the platform’s policy responses have proven insufficient. xAI maintains it did not build nudification tools and that users exploited the underlying model in ways the company neither intended nor anticipated. Minnesota’s legislature took the position that intent is irrelevant when the harm is as severe as nonconsensual intimate imagery and the platform has demonstrated it cannot prevent the conduct at scale.
Judge Frank made no ruling on which position is correct. He only ruled that xAI had not established, on a three-day deadline of the company’s own making, that harm was imminent enough to pause a law he had not yet fully examined. The restraining order is denied. The ban is live. Whether xAI can make its constitutional argument stick when August 19 arrives is the one question no court has answered yet.
