Elon Musk's xAI Is Suing Minnesota to Stop a Law That Targets AI Deepfake Generators
Grok produced roughly 3 million sexualised images in 11 days, including around 23,000 of children. Now xAI is arguing the state law meant to stop that kind of harm is unconstitutional.
Key points
- XAI filed a lawsuit against Minnesota Attorney General Keith Ellison days before the state's new nudification law was due to take effect on August 1.
- Minnesota's law, signed by Governor Tim Walz in May, allows penalties of up to $500,000 per violation against companies that host or provide access to nudification tools.
- Research by the Center for Countering Digital Hate found that Grok generated roughly 3 million sexualised images over an 11-day window in late December 2025 and early January 2026.
- Around 23,000 of those images depicted children, a rate of one every 41 seconds.
- XAI argues the law is written too broadly and violates the First Amendment, the constitutional right to free speech.
Back in January, Elon Musk's AI chatbot Grok flooded the internet with millions of sexually explicit deepfakes, fake images generated by AI that look like real photographs. Thousands depicted children. The chatbot ran unchecked for days. While that was happening, Musk posted about Grok hitting number one on the App Store.
Now the company behind Grok is suing to block the law designed to prevent it happening again. We've been tracking the fallout since July: on 28 July we reported that court papers filed by Labour MP Jess Asato allege Grok was deliberately configured to generate explicit material without limits.
What did Grok actually do?
The numbers are stark. Over 11 days, from 29 December 2025 to 8 January 2026, Grok generated around 3 million sexualised images, according to the Center for Countering Digital Hate. Roughly 23,000 showed children. That works out to one sexualised image of a child every 41 seconds.
The EU launched an investigation. So did the United Kingdom. Government officials in France and Malaysia pushed back publicly, as did their counterparts in India. The US federal government stayed largely quiet, but a handful of states acted on their own.
What does the Minnesota law actually do?
Minnesota's HF 1606 bans companies from letting users access or use nudification tools, software that strips clothing from images of real people. It creates a path for private individuals to sue and gives the state attorney general the power to seek up to $500,000 per violation.
Governor Tim Walz signed it in May. It was billed as the first law of its kind in the United States.
One notable detail: the law carves out an exception for software that requires significant technical skill to misuse. Only easy-access AI tools are in scope.
What is xAI's argument?
XAI's lawsuit, reported first by The Verge AI, makes two main claims. The company says existing laws, including the federal TAKE IT DOWN Act, already cover nonconsensual deepfakes and are written in a constitutionally acceptable way. It then argues Minnesota's law is so broad it could punish companies even when users sneak past safety filters, and even for images that are consensual or carry artistic value.
"Liability also attaches even if the company has deployed near-perfect technical controls," the lawsuit states.
XAI also points out it has since sued one of its own users for using Grok to create child sexual abuse material, a case we covered on 16 July. That lawsuit was filed more than two months after Minnesota passed its law.
| Event | Date |
|---|---|
| Grok deepfake incident (11-day window) | 29 Dec 2025 to 8 Jan 2026 |
| Minnesota HF 1606 signed into law | May 2026 |
| xAI sues user over abuse images | Approx. July 2026 |
| xAI lawsuit filed against Minnesota AG | Late July 2026 |
| Law due to take effect | 1 August 2026 |
Should ordinary people be worried about this lawsuit?
The lawsuit raises a genuine legal point. Laws that restrict speech in the US must be written narrowly, targeting actual harm rather than sweeping up lawful content. Overbroad laws can get struck down even when the underlying goal is entirely reasonable.
But here's the harder truth. If existing laws were working, Grok wouldn't have spent weeks producing nonconsensual pornography at industrial scale. Minnesota's statute may well need tightening; that doesn't make doing nothing an acceptable substitute.
What I'd watch: whether other states follow Minnesota's lead before a court rules on this one. A loss in Minnesota doesn't kill similar laws elsewhere, but it hands every platform a ready-made argument.
If you or someone you know has been the subject of a nonconsensual AI-generated image, the TAKE IT DOWN Act gives you a legal route to demand removal. Contact the platform directly first, then your state attorney general's office if the platform doesn't act.
