Minnesota's AI nudification ban could outlaw swimsuit photos and political satire, lawyers argue

A libertarian law firm is backing Elon Musk's xAI in a federal challenge to Minnesota's sweeping new ban on AI-generated "nudification", a law so broadly written, the challengers say, it could punish someone for creating an image of Wonder Woman or a man in a Speedo.

The Liberty Justice Center filed a friend-of-the-court brief Friday in U.S. District Court supporting xAI's lawsuit against Minnesota Attorney General Keith Ellison over HF 1606, a state law that took effect August 1. The law bans AI-generated imagery that depicts a person's "intimate part", a term it borrows from a criminal statute on sexual contact and defines to include the inner thigh, buttocks, or breast. Violations carry fines up to $500,000 each, enforceable by private litigants.

That definition, xAI and its allies argue, turns routine images into potential violations. And the law's lack of any intent requirement, safe harbor for good-faith removal, or exception for art, satire, journalism, or political speech makes it one of the broadest speech restrictions any state has attempted in the AI era.

$500,000 fines for swimsuits and superheroes

xAI's complaint, filed less than two weeks before the law took effect, laid out a series of scenarios that illustrate the statute's reach. The company argued in its filing that the law's ban on depicting an "inner thigh, buttocks, or breast" would ensnare it for generating what it called:

"benign images depicting individuals in shorts (inner thigh), men without shirts (breast), and competitive swimmers in speedos (buttocks and breast)."

The Liberty Justice Center's amicus brief went further. It warned that the law would expose xAI to a $500,000 fine if a user depicted himself as a partially clothed superhero like Wonder Woman or the Incredible Hulk, even if the image never left the user's own computer.

The brief called the law "far more egregious" than a federal ban on depictions of animal cruelty that the U.S. Supreme Court struck down. The center, best known for its Supreme Court victory against public-sector unions collecting mandatory fees from non-members, has also backed The Babylon Bee in a successful challenge to a California law banning "materially deceptive" memes and parodies of elected officials.

In the amicus filing, the center framed the stakes in broad First Amendment terms:

"Regardless of its motivations, Minnesota's heavy-handed attempt to restrict speech imperils the protected right of millions of Americans to use AI technology to express themselves."

Trump's Reflecting Pool image lands in the legal crosshairs

xAI's legal memorandum included a specific example designed to show the law's absurdity in action: an AI-generated image that President Trump shared on social media depicting himself, Vice President JD Vance, Secretary of State Marco Rubio, Interior Secretary Doug Burgum, and an unidentified, possibly fictional, woman lounging in the Reflecting Pool, all wearing sex-specific swimwear.

xAI called the image "political satire of a familiar kind." But under HF 1606, the company argued, the image unlawfully "nudifies" identifiable individuals because each man's breast is depicted where it was not depicted before. The law draws no distinction between revenge porn and a lighthearted political meme, and it imposes strict liability regardless of whether anyone was harmed or whether the image was ever shared publicly.

That lack of distinction is by design. State Sen. Maye Quade, the Democratic-Farmer-Labor sponsor of HF 1606, said she "intentional[ly]" left out an exemption for consensual images. She also sponsored a separate Minnesota election deepfake law. Federal courts have already invalidated similar election deepfake statutes in California and Hawaii, and a Minnesota appeals court kept the state's own version alive only by reading an unwritten parody exception into it.

Minnesota's track record on speech-adjacent regulation has drawn repeated legal challenges in federal court. The state's outright ban on prediction markets was blocked by a judge before it could take effect this month.

Ellison's opposition dodged the law's broadest language

AG Ellison filed an opposition to xAI's request for an emergency temporary restraining order before the law took effect. In that filing, Ellison quoted xAI's own admission that Minnesota has an interest in "curbing the non-consensual dissemination of artificially generated nude images."

But xAI's lawyers noted that Ellison's opposition "conspicuously omit[ted]" the law's sweeping definition of "intimate part" and its lack of any safe harbor for platforms that act in good faith to remove offending content. The attorney general's brief, in other words, defended the law by avoiding the provisions that make it most constitutionally vulnerable.

xAI itself already prohibits the kind of content the law targets. The company's Acceptable Use Policy, dated January 2, 2025, banned "[d]epicting likenesses of persons in a pornographic manner." A revised version issued June 26 added a prohibition on "[u]ndressing or nudifying real persons, or otherwise altering a real person's image or likeness to depict them in an intimate or sexual context."

In its memorandum, xAI stated plainly that it "strictly prohibits its users from generating nude or sexualized images of people without their consent." The company's objection is not to the goal of preventing revenge porn, it is to a law that treats anyone "who edits a photograph of herself" no differently than "a stranger who edits her picture," and that punishes xAI for "an image seen by no one and deleted at once."

President Trump himself signed bipartisan legislation last year criminalizing users who create revenge porn, including AI-altered images, and requiring platforms to remove such content within 48 hours of a valid request. That federal law, xAI's filings suggest, already addresses the harm Minnesota claims to target, without the constitutional overreach that HF 1606 introduces.

Judge Frank denied the emergency block, but left the door open

U.S. District Judge Donovan Frank, a President Clinton nominee managing a reduced caseload because of senior status, denied xAI's emergency TRO before August 1. He faulted the company for waiting nearly three months after Governor Tim Walz signed the law and filing "only three days before the law is set to take effect."

Frank did not, however, rule on the merits. He ignored Ellison's substantive arguments and instead scheduled an August 19 emergency hearing on xAI's motion for a preliminary injunction. That hearing will be the first real test of whether HF 1606 can survive constitutional scrutiny.

The procedural posture matters. xAI may have hurt its own cause by waiting so long to file. But the underlying question, whether a state can impose strict liability and half-million-dollar fines on an AI platform for generating images of people in shorts, does not depend on the plaintiff's litigation timing. Courts across the country have been increasingly willing to block government overreach when First Amendment rights are at stake.

xAI did not respond to queries from the reporter who broke the story. The company's silence leaves its legal filings to speak for themselves, and those filings paint a picture of a law drafted with a narrow problem in mind but written so broadly that it criminalizes political cartoons, fitness photos, and comic-book costumes.

NPR and a Minnesota NPR affiliate covered HF 1606 but went out of their way to omit the law's sweeping definition of "intimate part," even while highlighting its potential application to satirical images like Trump's Reflecting Pool post. That selective framing left their audiences with the impression that the law targets only malicious deepfakes, not the vast universe of ordinary images it actually reaches.

The Liberty Justice Center's brief put the core problem plainly: the law "prohibits vast quantities of expressive speech that has nothing to do with protecting privacy." It requires no intent, no knowledge, no evidence that an image was shared. It lacks the kind of narrowing provisions that courts have demanded when government restricts speech, and it duplicates narrower federal statutes that already address the same harm.

Protecting victims of AI-generated revenge porn is a legitimate goal. But when a law is so carelessly written that it turns a shirtless jogger into a nudification victim and a political meme into a half-million-dollar offense, the problem is not artificial intelligence, it is the people who drafted the statute.