August 1, 2026
A Minnesota judge has denied xAI's request to block a state law banning apps that let users generate nonconsensual nude images -- so-called "nudify" tools -- from operating in the state, according to TechCrunch. xAI sued to stop the ban, and lost, at least for now. That matters well beyond Minnesota. It's a signal that courts are willing to let states regulate specific, harmful AI outputs even when a company frames the fight as a free-speech or interstate-commerce issue. Other states watching this ruling now have a template, and other AI companies building anything adjacent to image generation should assume more of these laws are coming, not fewer.
For business buyers, the lesson isn't really about nudify apps specifically -- it's about the regulatory pattern. AI vendors are going to keep testing the limits of what states can restrict, and losing some of those fights. If you're building or buying AI-powered tools that touch user-generated images, likenesses, or anything with an obvious abuse vector, don't assume the current legal landscape is settled. It isn't. Vendors who can show they've thought about misuse before regulators force the issue will have an easier time keeping enterprise customers than ones who wait for a lawsuit to tell them where the line is. That's part of why we've been tracking how security and governance show up as buying criteria, not afterthoughts, across the AI tool market.
YouTuber Hank Green posted a striking admission this week: his own AI usage, he said, is "not healthy," and the dopamine he gets from interacting with LLMs isn't good for him or, in his words, for the world. TechCrunch covered the apology, and it's worth taking seriously precisely because Green isn't an AI skeptic by trade -- he's someone who has spent years thinking publicly about technology and media literacy. When a thoughtful, tech-literate creator says he got hooked on chatting with a model, that's a data point businesses deploying AI assistants internally shouldn't wave away as a fringe concern.
Here's my take: the AI industry has spent two years selling engagement as a feature -- assistants that remember you, chat naturally, and are available at 2 a.m. Those are the same design choices that made social media compulsive, just pointed at a more capable and more persuasive product. Businesses rolling out AI copilots to employees or customers should ask a question most vendors don't want asked: is this tool designed to be useful, or designed to be sticky? Those aren't the same thing, and conflating them is how you end up with a workforce that trusts a chatbot's judgment more than it should, or a support tool that keeps customers talking instead of resolving their problem. If you're evaluating an AI-powered helpdesk or CRM layer, it's worth asking the vendor directly how they think about usage patterns and whether the product is optimized for resolution or retention.
Put these two stories together and a theme emerges: the guardrails around AI -- legal and personal -- are being built in public, in real time, mostly by people who got burned first. Minnesota's law came from documented harm. Green's apology came from noticing his own behavior before it went further. Neither of these is an argument against using AI tools in your business. They're an argument for building in judgment about where the line is, rather than assuming the vendor has already drawn it for you. Companies that treat AI adoption as a values decision as much as a productivity decision are going to end up with fewer headaches down the road.
Where's your line -- have you noticed your own team (or yourself) leaning on an AI tool more than feels healthy, and did you do anything about it?
Sources