xAI sued Minnesota on Monday to block a state law that would fine AI image tools up to $500,000 for every nudified output discovered in the state, arguing the statute is unconstitutional and would force Grok Imagine to strip out its image-editing features when the law takes effect August 1, 2026. Elon Musk's company says the penalty structure creates strict liability with no safe harbor, exposing xAI to civil damages that scale into the billions once even a small fraction of its user base misuses the product. The complaint arrives after more than six months of backlash and probes over Grok-generated sexual imagery, and days after xAI was sued by a 10-year-old Jane Doe in Arkansas whose photographer allegedly used the tool to sexualize client images.
The math in xAI's filing is the argument. Ten violative images would generate $5 million in civil penalties. A thousand would reach $500 million. A hundred thousand — a threshold xAI calls plausible for a product with millions of users generating billions of images — would produce $50 billion in exposure. Minnesota's law also grants a private right of action, meaning victims can sue over individual outputs on top of state penalties.
xAI's fallback position is that it is already policing bad actors. The company said it has received 70,000 reports of suspected Grok-generated child sexual abuse material and that at least 244 of those reports have led to arrests. Earlier this month, xAI began suing users it accuses of circumventing Grok's safeguards, a maneuver the new Minnesota complaint holds up as evidence the company deserves safe harbor rather than direct liability.
Key facts
- 01xAI sued Minnesota on Monday to block a nudify-app ban that fines AI firms up to $500,000 per harmful image starting August 1, 2026.
- 02xAI says 10 violative images would trigger $5M in penalties, 1,000 would reach $500M, and 100,000 could hit $50B.
- 03The suit follows 70,000 reports of suspected Grok-generated CSAM, which xAI says led to 244 arrests.
- 04A 10-year-old Jane Doe alleges an Arkansas photographer used Grok to generate hundreds of images of her and thousands of images of other children.
- 05xAI is invoking the First Amendment and pointing to the federal Take It Down Act as a less restrictive alternative.
The company frames the enforcement risk in operational terms rather than moral ones. xAI says it would prefer to leave Grok Imagine's editing feature exactly as it operates today and rely on terms of service that ban CSAM and non-consensual intimate imagery. The $500,000-per-image penalty, it argues, makes that posture financially impossible.
“Confronted with $500,000-per-image strict liability and no safe harbor, xAI has no practical choice but to restrict Grok Imagine's image-editing features in various ways when the statute takes effect on August 1, 2026.”— xAI, from the company's complaint
The First Amendment argument sits at the core of the complaint. xAI says Minnesota's law sweeps in nude images with artistic, scientific, satirical, educational, medical, or religious value, and imposes liability even when the depicted person consented, created the image themselves, or never shared it. The company argues the federal Take It Down Act, which criminalizes distribution of real and AI-generated intimate depictions, already covers the actual harm Minnesota claims to address.
Much of the complaint turns on how Minnesota defines intimate parts. The state pulled its definition from a criminal statute governing nonconsensual touching, which covers inner thighs, buttocks, and male or female breasts. xAI argues that standard bans ordinary depictions of shirtless men, people in shorts, and swimsuit imagery, and points to political satire — reimagining politicians in Speedos or as Sumo wrestlers — as protected expression that would be chilled. Notably, the filing largely sidesteps the bikini-generation feature Musk himself promoted on the X app, which drove much of the initial backlash.
Minnesota Attorney General Keith Ellison signaled the state will defend the law. His office said it is waiting to be served and will respond in court.
“But I know that using AI to generate nude images of people against their will is appalling. There are plenty of worthy debates to have about AI policy. This is not one of them.”— Keith Ellison, Minnesota Attorney General
The Jane Doe case shows why the state passed the statute. Her lawyer told Ars Technica that many of the other children the Arkansas photographer allegedly targeted are preparing their own suits, which could be folded into a potential class action filed by young girls in Tennessee in March. The photographer allegedly generated hundreds of images of Doe and thousands of images and videos of other clients using Grok.
xAI's legal theory — that chatbot outputs are protected expression comparable to human speech — is the same argument other AI providers have deployed in adjacent litigation over defamation and harmful content. Whether it survives depends on whether the court agrees Minnesota's law fails narrow-tailoring under strict scrutiny. xAI argues it fails any level of scrutiny.
The counter-argument the state is likely to press is that Minnesota is regulating conduct at the point of image creation because distribution-focused laws like the Take It Down Act arrive after the harm is already done. If the court sides with xAI and enjoins the law, other states with pending nudification statutes will face the same First Amendment ceiling. If Minnesota prevails, image-generation providers may have to geofence features, tighten prompt filtering, or exit the state — outcomes xAI's complaint explicitly warns against.
The case matters beyond Grok because it is the first serious constitutional test of state-level AI liability written as strict per-image penalties. Every frontier image model — from OpenAI's Sora to Google's Imagen and Meta's Emu — ships with editing features that a determined user can push toward NCII outputs, and every provider has been relying on the same defense-in-depth of terms of service, content filters, and post-hoc user bans. A $500,000-per-image regime does not permit defense in depth; it requires prevention that approaches zero-failure, which no current model can guarantee. If Minnesota's approach survives, the compliance cost of running a public image tool in the US goes up sharply, and the incentive shifts toward heavier pre-filtering, tighter API gating, and outright feature removal in flagged jurisdictions. That is the market outcome xAI is trying to prevent, and it is the outcome plaintiffs' lawyers in Tennessee and Arkansas are counting on.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
