xAI has sued Minnesota Attorney General Keith Ellison to block a state law that would expose AI platforms to penalties of at least $500,000 for each unlawful use of “nudification” technology. The law takes effect August 1, 2026, putting Grok Imagine and other image-generation services on a collision course with a state-level attempt to stop nonconsensual synthetic sexual imagery at the point of creation.
As first reported by the Minnesota Reformer and confirmed in reporting by the Associated Press, Elon Musk’s AI company argues that Minnesota’s statute is an unconstitutional, content-based restriction on speech. xAI is seeking a federal declaration that the measure is invalid and an injunction preventing Ellison from enforcing it.
Minnesota’s law, HF 1606, makes it unlawful for the owner or controller of a website, app, or program to let a user access, download, or use a service to nudify an image or video. The statute defines that as generating or altering media to depict an intimate part not present in an original image of an identifiable person, where the result looks realistic enough that a reasonable person would believe it belongs to that individual.
Minnesota already has laws addressing the nonconsensual dissemination of intimate imagery. This new statute reaches earlier in the chain: it targets providers that make the generation tool available to Minnesotans, whether or not the resulting image is ultimately posted or shared.
That distinction is central to xAI’s challenge. The company says it does not oppose prohibitions on distributing AI-generated nude images of real people without consent, but argues that Minnesota’s approach sweeps in protected artistic, political, scientific, and satirical work. Its complaint contends that the statutory definition of “intimate part” is insufficiently precise and could capture innocuous images involving bathing suits, shirtless men, or other nonsexual depictions.
The Minnesota Revisor’s published text makes the scope especially consequential for cloud AI vendors: a separate unlawful “access, download, or use” can trigger the statutory penalty. For a consumer service with high-volume image generation, the exposure is not limited to a single moderator failure or a single published deepfake.
Those assertions may help xAI argue that its service has existing guardrails and enforcement processes. But the statute is structured around availability of the functionality itself, not simply whether a provider’s terms prohibit misuse. That creates a difficult compliance question for generative-AI operators: can a service rely on filters, user rules, reporting, and account enforcement, or must it technically prevent Minnesota users from accessing any capability that could be used to create a realistic synthetic nude?
Attorney General Ellison has said his office had not yet been served when contacted by reporters, but rejected the premise that this is a marginal AI-policy issue. He described AI nudification as conduct that strips targets of dignity and can cause emotional, personal, and professional harm. Rep. Jessica Hanson, a chief author of the measure, said victims are overwhelmingly women and children and criticized xAI for fighting a law intended to curb sexual exploitation.
The broader implications could be substantial. If Minnesota can impose provider-side liability for every in-state use of a generative imaging feature, AI companies may respond with stronger model-level restrictions, regional feature blocks, identity or location checks, or a more conservative approach to image editing across the United States.
The law is scheduled to take effect Saturday, August 1. Unless the federal court grants xAI emergency relief, Minnesota’s enforcement framework will arrive before the First Amendment challenge is resolved.
As first reported by the Minnesota Reformer and confirmed in reporting by the Associated Press, Elon Musk’s AI company argues that Minnesota’s statute is an unconstitutional, content-based restriction on speech. xAI is seeking a federal declaration that the measure is invalid and an injunction preventing Ellison from enforcing it.
Minnesota’s law, HF 1606, makes it unlawful for the owner or controller of a website, app, or program to let a user access, download, or use a service to nudify an image or video. The statute defines that as generating or altering media to depict an intimate part not present in an original image of an identifiable person, where the result looks realistic enough that a reasonable person would believe it belongs to that individual.
The dispute is over platform liability, not just bad actors
Minnesota already has laws addressing the nonconsensual dissemination of intimate imagery. This new statute reaches earlier in the chain: it targets providers that make the generation tool available to Minnesotans, whether or not the resulting image is ultimately posted or shared.That distinction is central to xAI’s challenge. The company says it does not oppose prohibitions on distributing AI-generated nude images of real people without consent, but argues that Minnesota’s approach sweeps in protected artistic, political, scientific, and satirical work. Its complaint contends that the statutory definition of “intimate part” is insufficiently precise and could capture innocuous images involving bathing suits, shirtless men, or other nonsexual depictions.
The Minnesota Revisor’s published text makes the scope especially consequential for cloud AI vendors: a separate unlawful “access, download, or use” can trigger the statutory penalty. For a consumer service with high-volume image generation, the exposure is not limited to a single moderator failure or a single published deepfake.
Grok’s safeguards will be part of the factual fight
xAI says users of Grok and Grok Imagine must accept terms that prohibit child sexual abuse material and pornographic depictions of a person’s likeness. In its complaint, the company also says it has suspended more than 50,000 accounts, made more than 70,000 reports to the National Center for Missing & Exploited Children, and that those reports led to 244 arrests.Those assertions may help xAI argue that its service has existing guardrails and enforcement processes. But the statute is structured around availability of the functionality itself, not simply whether a provider’s terms prohibit misuse. That creates a difficult compliance question for generative-AI operators: can a service rely on filters, user rules, reporting, and account enforcement, or must it technically prevent Minnesota users from accessing any capability that could be used to create a realistic synthetic nude?
Attorney General Ellison has said his office had not yet been served when contacted by reporters, but rejected the premise that this is a marginal AI-policy issue. He described AI nudification as conduct that strips targets of dignity and can cause emotional, personal, and professional harm. Rep. Jessica Hanson, a chief author of the measure, said victims are overwhelmingly women and children and criticized xAI for fighting a law intended to curb sexual exploitation.
A test case for AI tools delivered through the browser
For Windows users, the immediate issue is not a change to Windows 11 or Microsoft’s AI stack. Grok is a cloud service accessed through the web and apps, so any ruling will turn on the provider’s controls, policies, and location-based service design rather than on a client PC’s operating system.The broader implications could be substantial. If Minnesota can impose provider-side liability for every in-state use of a generative imaging feature, AI companies may respond with stronger model-level restrictions, regional feature blocks, identity or location checks, or a more conservative approach to image editing across the United States.
The law is scheduled to take effect Saturday, August 1. Unless the federal court grants xAI emergency relief, Minnesota’s enforcement framework will arrive before the First Amendment challenge is resolved.