Minnesota’s deepfake law could expose AI companies to fines of up to $500,000 per image or video, even when they have safeguards against misuse. An appeal by xAI has put the law’s reach, and its potential effect on AI-generated speech, before the Eighth Circuit.
xAI Watch analysis
What happened
Lawfare’s Jeff Kosseff explains that Minnesota’s law allows the state attorney general to seek fines when a service lets a user create an image or video that depicts an identifiable person with an intimate part not present in the original. The law does not require proof that the company intended the use, or that the depicted person did not consent. Kosseff notes that the definition is broad enough to raise questions about images that are not intimate in the ordinary sense, such as a bathing-suit image showing someone’s inner thigh.
xAI, which develops Grok, sued Minnesota and sought to block the law. A district court denied its request for a preliminary injunction, focusing on the time xAI took to bring the case rather than deciding the constitutional questions. On 2 October, the Eighth Circuit temporarily barred enforcement of the law against xAI while it considers the appeal. That is an interim measure, not a final ruling on whether the law is constitutional.
Why it matters
The case tests whether a state can hold an AI provider responsible for users’ image generation without requiring intent, recklessness or knowledge. That matters beyond deepfakes: rules aimed at harmful uses of AI may also affect what services can offer and what people can create.
There is a real harm on one side of this dispute and a difficult question about the breadth of the remedy on the other. The eventual ruling could help define how courts balance protection from abusive deepfakes against restrictions that may reach consensual images, satire or other protected expression.
Our read
The useful distinction here is between the goal of stopping abuse and the means chosen to do it. A fine that can apply despite safeguards makes the standard of liability central, not a footnote. The appeal is worth following, but the temporary injunction should not be mistaken for a final verdict. Court cases are rather good at making that distinction matter.
What to watch
- Whether the Eighth Circuit addresses the law’s strict-liability standard.
- How the court treats the law’s scope, including consent and protected expression.
- Whether the eventual decision affects AI providers beyond xAI and Minnesota.
Discussion spark: Should AI companies face liability when a user bypasses safeguards to make a prohibited image, or should the law require proof that the company knew about or intended the misuse?
Sources and evidence
- Source update (4 November 2026, 00:00 UTC)
- Delhi HC orders takedown of Kriti Sanon deepfakes, impersonation content; says 'personality rights' cannot justify blanket injunction | Delhi News – The Times of India (4 November 2026, 00:00 UTC)
not affiliated with or endorsed by xAI, SpaceXAI, SpaceX, X Corp. or Cursor