If you are a model, a photographer, or a designer who posts real people wearing your clothes, the Minnesota nudification fight is not an abstract Elon Musk story. It is about who bears the cost when a stranger runs your campaign image through an AI tool built to strip clothing off a real body. On September 4, a federal judge in Minnesota denied xAI's request to block the state's nudification ban while the company's constitutional challenge plays out. That is new since we last checked in on this case in August, and it leaves the first-in-the-nation law fully enforceable — though for reasons that have less to do with the First Amendment than you might expect.

The ruling

Minnesota's HF 1606 took effect August 1, 2026, and bars anyone who owns or controls a website, app, or service from letting users nudify an image of a real, identifiable person. xAI sued to stop it, first seeking a temporary restraining order, which U.S. District Judge Donovan W. Frank denied on July 31, faulting the company's delay. He set the matter for an expedited hearing on August 19 as a motion for preliminary injunction instead. That motion has now been denied too.

Read the opinion closely, because the headline and the holding are not the same thing. Frank did not rule that HF 1606 is constitutional. He ruled that xAI failed to carry its burden for emergency relief. The law was signed May 7; xAI filed nearly three months later and three days before the law took effect, offering no explanation beyond the observation that it takes time for a law to reach corporate management. Frank noted there is "no serious dispute that xAI is a sophisticated and well-resourced litigant with the means to move quickly when its interests are at stake," and concluded that if the company genuinely feared irreparable harm, it would have acted faster. He also found the harm xAI did allege — a monetary penalty, engineering time already spent, speculative lost users — was either recoverable or unproven. On that basis alone, he wrote, the motion "is properly denied."

Only then did he reach the balance of harms and the public interest, which he found "tips steeply in favor of the State" — a statute passed by the House in April, unanimously by the Senate, and aimed at a harm xAI itself does not dispute. The constitutional question is expressly reserved: the State has a pending motion to dismiss that puts the First Amendment issues "front and center," and Frank said those issues "deserve, and will receive, full consideration."

Why it matters

Three features of HF 1606 are worth understanding if your work involves images of real people.

There is no consent defense. The word "consent" does not appear in the statute. Liability does not turn on whether the person depicted objected, agreed, or made the image themselves. During a Senate hearing, when a staffer pointed out that the prohibition reaches consensual images, the bill's principal sponsor said that was intentional. For a model who has signed off on AI-assisted retouching, and for the brand that commissioned it, that is a different risk picture than the consent fights in the Rainbow Shops or EBY disputes.

"Intimate part" is broader than nudity. The law borrows its definition from Minnesota's criminal sexual contact statute: the primary genital area, groin, inner thigh, buttocks, or breast. An altered image is covered if it depicts one of those areas where the original did not, and is realistic enough that a reasonable person would believe the body is the depicted individual's. Swimwear, lingerie, and activewear imagery sits closer to that line than most people assume — a point xAI presses hard in its complaint, which offers shirtless politicians and a viral sumo-wrestling meme as examples of what the text sweeps in.

There is one exemption, and it is narrower than a safe harbor. The prohibitions do not apply where the service "requires the technical skill of a user to nudify" — defined as substantial individualized technological or artistic skill and judgment by a human directing the output. That is what separates a professional editing suite from a one-prompt generator. It is not a defense for good-faith filtering. xAI's core complaint is that the statute has no scienter requirement and no credit for terms of service, moderation, or state-of-the-art technical controls — a strict-liability theory the court has not yet accepted or rejected.

The penalty structure is what has already changed behavior. The attorney general can seek civil penalties of up to $500,000 for each unlawful access, download, or use, and the law separately gives any individual depicted a private cause of action for treble damages, punitive damages, injunctive relief, and attorney's fees. That second piece is the one independent models and creators should note: the statute does not require you to wait for the attorney general.

xAI restricted Grok Imagine's image-editing features for Minnesota users on August 1, and told the court its moderation rate on image-editing prompts in the state has risen more than 10 percent. The court's own summary was blunter: xAI "has disabled its nudification tool in Minnesota." That is a company changing its product because a state law has teeth, before any court has found the law constitutional. Frank then used that compliance against it — the adjustment was already made, so no future cost was shown.

What to watch

The merits. The State's motion to dismiss is the vehicle that will actually test whether HF 1606 is a content-based restriction that fails strict scrutiny, and Frank signaled a ruling in the near future. That is the decision other states will read.

An appeal is likely but narrower than it sounds. Reporting indicates xAI has said it will take the denial to the Eighth Circuit. Worth keeping in perspective: an appeal from a denied preliminary injunction reviews whether the district judge abused his discretion, and Frank rested on delay and irreparable harm, not on the First Amendment. An affirmance would say little about whether the statute survives.

The private suits keep coming. The court's order cites three pending cases alleging that photographs of children were used to create child sexual abuse material on Grok Imagine — one in the Northern District of California and two in the Eastern District of Arkansas. Separately, a Labour MP filed a claim in the UK High Court in June over nonconsensual deepfakes. The litigation front is widening independent of Minnesota.

For fashion and beauty brands using AI image tools, the practical takeaway is not to wait for the Eighth Circuit. If a vendor's tool can alter a real person's image, the question to put in writing is how it prevents nudification — not whether its terms of service prohibit it. Under HF 1606, the terms of service are not the answer.

Sources