Minnesota has drawn a line most states have avoided: it is going after the companies that build “nudify” apps, not only the people who use them. The state’s nudification ban took effect on August 1, and days earlier a federal judge refused to block it, letting the law stand against a challenge from Elon Musk’s xAI.
The shift in target is the whole point. Laws against deepfake nudes usually punish the person who made or shared an image. Minnesota’s HF 1606 reaches upstream to the tool itself, which the state describes as the first time a U.S. law puts direct liability on the AI companies that make nudification possible.
What the law does
The mechanism is money, at scale. HF 1606 sets a civil penalty of up to $500,000 for each unlawful access, download, or use of a covered nudify tool. Aimed at makers rather than only end users, that is a structure meant to make offering the tool in Minnesota a serious financial risk.
It is not unlimited. The statute exempts tools that require specialized technical skill and preserves Section 230 protections, according to the reporting, which are the kind of carve-outs that will shape how far it reaches. Those limits are also where the legal fight will focus.
Why xAI lost the first round
Musk’s xAI moved to stop the law with an emergency request for a temporary restraining order. On July 31, Judge Donovan Frank of the District of Minnesota denied it. His reasoning leaned on timing: xAI filed its challenge nearly three months after the law was signed and only three days before it took effect, and the judge wrote that the delay “suggests that harm is not immediate.”
That is a procedural loss, not a ruling on the merits. The real test, whether a strict-liability law aimed at a tool survives a First Amendment challenge, is reported to come at a later preliminary-injunction hearing, though that date rests on a single outlet and is not firm.
The reach into the image economy
This lands close to the adult-content world even though it is framed as a safety law. Nudify tools sit at the ugly edge of AI image generation, producing fake explicit images of real people without consent, and a strict-liability rule against the companies that build them redraws who carries the risk. If holding a tool-maker liable survives court, the model spreads, and every company shipping image-generation features has to weigh what its tool can be pointed at.
The First Amendment question underneath is real, and one Minnesota ruling does not settle it. But the direction is set. The law is moving from chasing individuals who misuse a tool toward the companies that put the tool in their hands, and the outcome in Minnesota is the early read on whether that approach holds.
