Eighth Circuit Pauses Minnesota's First-in-the-Nation AI Nudify Ban
The Eighth Circuit has temporarily paused Minnesota's ban on AI nudification apps while xAI's First Amendment lawsuit proceeds. The law, the first in the nation to target platforms that generate fake nude images, took effect on August 1 after a district judge rejected xAI's injunction request. Enforcement now stays on hold until the appeal is decided.
On this page
A stay, not a ruling
The Eighth Circuit Court of Appeals has temporarily blocked enforcement of Minnesota's ban on AI nudification applications while the constitutional challenge from Elon Musk's xAI moves forward, according to wire reports on 2 and 3 October. The order, granted as an injunction pending appeal, does not strike the law down; it freezes enforcement until the appeals court can decide whether the statute survives First Amendment scrutiny.
The underlying law, House File 1606, took effect on 1 August as the first in the country aimed specifically at nudify platforms: tools that take photos of real people and generate fake nude or sexualized images from them. It punishes platforms that let users create such imagery and gives depicted people a path to sue. xAI sued Attorney General Keith Ellison in July, arguing the Constitution protects its ability to build and offer Grok, which users had already used to undress photos of real people before the company added blockers.
The First Amendment argument over Grok
The district court was not persuaded the first time. Judge Donovan Frank rejected xAI's request for a temporary restraining order on 31 July and its preliminary injunction on 4 September, letting the ban take effect while the case proceeded. xAI escalated on 11 September, asking the Eighth Circuit to enjoin the law pending its appeal, and the court granted that request three weeks later.
The fight turns on where building a tool ends and publishing content begins. xAI's position is that offering a general-purpose model is itself expressive activity, so a law that restricts offering Grok restricts speech. The state's position is that the statute regulates conduct, facilitating the creation of nonconsensual sexual imagery of real people, which has never enjoyed First Amendment protection, and that the people depicted did not consent to any of it. The appeals court has not resolved that question; it has only decided that xAI is likely to suffer irreparable harm if enforcement continues while the argument is heard.
What it means for the states writing AI laws
The stay lands in a policy environment running on state law. With Congress largely stationary, states have spent 2026 writing their own AI rules: Minnesota's nudify ban, California's no-robo-bosses employment statute and kill-switch executive order, and a stack of deepfake and disclosure laws. Every one of those frameworks now carries the same implicit warning this order illustrates: a single federal court can suspend a state's AI policy on constitutional grounds before most residents have read it.
For builders, two practical notes. First, an injunction pending appeal is temporary geography, not permission: the law remains on the books in Minnesota, and a final ruling could reinstate it with retroactive effect for conduct after the stay lifts. Second, the theory xAI is advancing, that the model builder is shielded because the tool is speech, is being tested on the most unsympathetic facts available, nonconsensual imagery of real people. Whatever the Eighth Circuit decides here will shape how easily the same argument transfers to platforms whose outputs cause other kinds of harm, and states drafting the next round of AI laws are watching that exactly as closely as the companies are.