xAI sues to block Minnesota's Aug 1 AI-nudify liability law
It fines platforms up to $500K per image with no safe harbor for safeguards — a First Amendment fight over whether a state can outlaw a model capability.
Copy markdown
Strict liability, no safe harbor — the part that hits builders
Minnesota's HF 1606 makes a platform liable up to $500,000 per image for user-generated nude fakes, and there's no safe harbor: banning the use, filtering prompts, and adding guardrails doesn't get you off the hook. If you host image generation, “we tried to stop it” isn't a defense.
What xAI is actually arguing
xAI sued in federal court against AG Keith Ellison, calling the statute an overbroad, content-based speech restriction that's “presumptively unconstitutional” and not narrowly tailored to a compelling interest. It wants the law declared unconstitutional and enforcement blocked.
The clock: it takes effect Aug 1
The law goes live August 1, 2026 — days away — so a ruling on xAI's injunction bid could land fast. If no injunction issues, the AG and private victims can start suing platforms immediately.
Why the 'intimate part' definition matters to you
xAI says the definition is so broad it sweeps in shirtless men and people in swimwear, not just explicit deepfakes. How the court reads that scope tells every image and video builder where the legal line actually sits.
The precedent to watch
This is the first real test of whether a U.S. state can pin strict, safe-harbor-free liability on a model capability itself. The outcome shapes what generation tools you can legally ship — and whether providers preemptively strip features to dodge per-image fines.