xAI Sues Minnesota Over Law Targeting “Nudification” Apps

xAI has filed suit against Minnesota Attorney General Keith Ellison, challenging a law passed back in May that broadly targets apps for “undressing” people in photos. According to the company, the statute's punitive provisions leave it “no practical choice but to restrict image-editing features in Grok Imagine in one way or another.” The law, xAI insists, runs afoul of the First Amendment.

The backdrop to the suit is telling. In January 2026, Grok flooded the internet with millions of explicit deepfakes, including images of minors. The Center for Countering Digital Hate, after analyzing 11 days from December 29 to January 8, counted roughly 3 million sexualized images, of which some 23,000 involved children — one such picture every 41 seconds. Even so, the service wasn't taken offline for several days, and at the height of the scandal Elon Musk was delighting in the fact that Grok was topping the App Store in country after country.

The scandal drew reactions worldwide: the EU and the UK opened investigations, and officials in France, India, and Malaysia voiced outrage. In the US, federal authorities kept quieter, but “blue” states — California and Minnesota — took a hard line.

What the Law Prohibits

Minnesota's law (HF 1606) bars website owners from giving users the ability to “access, download, or use” such software. A curious exception is carved out for technologies “that require the user's technical skill” — meaning manual Photoshop work doesn't fall under the ban. The statute establishes a civil cause of action for private individuals and allows the state's attorney general to seek fines of up to $500,000 per violation. Signed by Governor Tim Walz in May, it is set to take effect on August 1.

The puzzle is why xAI waited three months and filed suit only a couple of days before that deadline — as if someone at the company had suddenly grasped the looming violation and was scrambling to find lawyers willing to work overtime for a deepfake generator.

The Arguments and Their Weak Spot

The suit itself, though, looks quite reasonable. xAI points out that the law is overly broad: liability attaches even to images created with consent, to “mildly risqué” pictures of adults, or to material with humorous and artistic value. The company stresses that it applies cutting-edge technical safeguards and that violating them is only possible by circumventing its protections, and it also notes that it has already sued a user who generated CSAM (albeit more than two months after the law was passed).

At the same time, xAI's lawyers maintain that existing rules — including the TAKE IT DOWN Act — already prohibit the distribution of such deepfakes and are written in a constitutionally sound way. But this is precisely where the central contradiction lies: if the existing laws work, then where did the January torrent of pornography come from? Perhaps Minnesota's law really is poorly drafted. But what good are “good” laws if they still fail to make anyone behave properly?

Source: The Verge