Minnesota Moves to Ban AI Apps That Generate Fake Nude Images

In brief

• Minnesota lawmakers passed a bill banning AI tools that generate fake nude images.
• Violators face up to $500,000 per use and potential triple damages.
• The law keeps Section 230 protections and will takes effect August 1 if signed into law.

Minnesota lawmakers have passed a bill aimed at stopping a growing form of AI abuse by targeting the platforms that enable it. On Thursday, the Minnesota Senate voted 65-0 to pass House File 1606, sending it to Governor Tim Walz for his signature. The measure bars websites and apps from offering tools that generate realistic fake nude images of identifiable people. Under the bill, companies that control a website, app, or software service cannot allow users to access or use tools to create these images or generate them on a user’s behalf. Advertising or promoting such services is also prohibited.

The measure allows victims to sue the people or companies that operate or control nudification tools, such as websites, apps, or software that generate fake nude images. People depicted in AI-generated nude images can seek damages, including for mental anguish, and courts can award up to three times the actual damages, along with punitive damages, attorney fees, and orders to stop the conduct.  The bill also gives the state attorney general the power to enforce the law, with civil penalties of up to $500,000 per use. According to the bill, those penalties are directed into the state’s general fund and then appropriated to victim services, including support for survivors of sexual assault, domestic violence, and child abuse. The bill targets tools that require little technical expertise and have become widely accessible, including to minors. If signed, the law takes effect August 1 and applies to new cases from that date forward.

While the new bill doesn’t reference a single AI developer, the news comes after a series of high-profile incidents on the social platform X, including in August 2025, when Elon Musk’s xAI tool, Grok, generated nude deepfakes of Taylor Swift. The pop superstar moved to trademark her voice and likeness with the U.S. Patent Office in April, perhaps in a move to head off future AI reproductions. Musk is also facing mounting legal pressure, including a federal class action lawsuit filed by three Tennessee minors alleging Grok generated child sexual abuse material from their images. Also, a consumer protection lawsuit from the city of Baltimore claims the company knowingly deployed a system that produces and spreads nonconsensual sexualized content, including of minors. Public Citizen co-president Robert Weissman said the spread of these tools reflects how quickly AI has lowered the barrier to creating nonconsensual intimate imagery and expanded its reach. “These apps are 99% targeting women, over 90% of whom are under 18. It’s a tool of intimidation and harassment of women with really severe psychological consequences,” Weissman told Decrypt. “You’ve seen this across the country and the world. So the need for government intervention and regulation is acute.” Weissman added that state-level laws can play a role alongside federal efforts, especially when it comes to enforcement. He said local authorities may be better positioned to act quickly in individual cases, while federal agencies may not prioritize or pursue them at all. The Minnesota law also comes during an ongoing fight between President Donald Trump’s administration and states over who should control AI regulation. The Take It Down Act, signed into law by President Donald Trump in May 2025, criminalizes the distribution of nonconsensual intimate images and provides victims a path to seek civil damages. “I think having complementary federal and state standards is positive, particularly in theory. We’re talking about different enforcement systems and enforcement agencies,” Weissman said. “So you might have a federal standard, but you might not have federal capacity to do enforcement actions.” The office of Governor Walz did not immediately respond to Decrypt’s request for comment.