Minnesota's Nudification Ban Took Effect Today. xAI Asked a Judge to Stop It and Lost.
Bot Mutiny |
Minnesota is the first state to write a law aimed at the makers of nudification tools rather than the people using them. xAI filed 38 pages to block it three days before the deadline. A federal judge said no.
On Monday, lawyers for xAI filed 38 pages in federal court in Minnesota asking a judge to stop a law from taking effect.
The judge said no. The law took effect this morning.
Minnesota's HF 1606 bans nudification technology, the class of apps and websites that take an ordinary photograph of a real person and generate a fake nude image from it. Governor Tim Walz signed it in May. It is the first state law in the country aimed at the makers of the tools rather than the people using them, and that distinction is the entire fight.
What the law does that other laws do not
Most existing law in this area targets the user. Deepfake statutes in more than two dozen states penalize the person who creates or distributes a sexualized image of someone without their consent. The federal Take It Down Act, passed last year, requires websites and apps to build a removal process. Texas took a middle path, exposing a site or app owner to civil damages when the owner knew the depicted person had not consented, or failed to remove an image promptly after being told about it.
Minnesota went further up the chain. HF 1606 reaches the provider of the technology itself, and creates a path for both individual victims and the attorney general's office to bring an action. The penalty xAI cites in its complaint is $500,000 per violation.
That number is why the case exists.
xAI's argument
The complaint, filed against Attorney General Keith Ellison, does not defend nudification. xAI states that it does not contest Minnesota's interest in prohibiting the distribution of artificially generated nude images of real people without consent. Its position is that the statute extends far beyond that goal and imposes what the filing calls an "overbroad, content-based ban" on speech and on the tools of visual expression.
Three specific objections carry the argument.
The first is definitional. xAI argues the law does not adequately define "intimate part," and that as written it would reach images no ordinary person would describe as nudification: men without shirts, people in shorts, people in swimsuits. The complaint offers an example from the President's own social media account, an AI-generated image showing the president, the vice president, several cabinet members and an unidentified woman in bathing suits at the Lincoln Memorial Reflecting Pool. Bare-chested men, xAI argues, would fall inside the ban.
The second is the absence of a safe harbor. The complaint says the statute contains no provision protecting a company that makes good-faith efforts to stop users from generating prohibited images. xAI states that its terms of service already prohibit generating nude or sexualized images of people without consent, that it has suspended more than 50,000 accounts, and that it has made more than 70,000 reports to the National Center for Missing and Exploited Children, resulting in 244 arrests.
The third is arithmetic. At $500,000 per violation with no safe harbor, xAI's lawyers argue that a company whose users produced 100,000 prohibited images would face a figure they put at $50 billion.
What the judge said
U.S. District Judge Donovan Frank denied the request to pause the law. His reasoning was about timing rather than the First Amendment. The law was signed in May. xAI filed its motion on July 29, nearly three months later and three days before the effective date. Frank found little risk of immediate harm in letting the statute take effect as scheduled.
That is not a ruling on whether the law is constitutional. The case continues, and the constitutional question is genuinely open. xAI won a partial victory on a related theory in California, where a federal judge struck down a law restricting AI-generated deepfakes during elections. The company has also sued over California's training data transparency act, and Musk's X sued Minnesota separately over its election deepfake statute. This is a litigation strategy with a record behind it, not a one-off.
The proceeding the complaint does not mention
xAI's filing argues in part that Minnesota's law is unnecessary because the company already polices its own platform. That claim is being tested elsewhere, in a case the complaint does not raise.
The company faces a proposed class action from plaintiffs who allege that Grok was used to create and share sexual images built from real photographs of them, including images of them as children. The suit further alleges that xAI did not pass information about the alleged perpetrators to authorities. The 50,000 suspensions and the 70,000 reports to NCMEC are the company's own figures, offered in its own filing, and no court has yet weighed them against what the class action alleges.
Ellison's response was brief. He said he was waiting to be served, and that using AI to generate nude images of people against their will is appalling. Walz was briefer: "See you in court, creep."
What the record establishes, and what it does not
The record establishes that Minnesota passed the first state law in the country reaching the makers of nudification tools, that the law took effect on August 1, that xAI moved to block it on First Amendment grounds four days earlier, and that a federal judge declined to pause it on timing grounds.
It does not establish that the law is constitutional. It does not establish that xAI's overbreadth reading is wrong, and the swimsuit example is a real question about statutory drafting that deserves an answer rather than a dismissal. It does not establish anything about the truth of the class action's allegations, which remain allegations.
What the filings do show is where the argument has moved. For three years the fight over generated sexual images was about the person typing the prompt. Minnesota decided to ask what the company that built the box owes. The first company to answer arrived with a constitutional objection, a bill for $50 billion, and three days on the clock.
Sources: Minnesota HF 1606, signed May 2026; xAI's complaint against Attorney General Keith Ellison, filed July 27, 2026 in the U.S. District Court for the District of Minnesota; Judge Donovan Frank's July 30 order denying a temporary pause, as reported by NBC News; contemporaneous reporting from CBS Minnesota, CNBC, and the Minnesota Reformer.