The first state law that doesn't just punish the users, it goes after the toolmakers.
The Summary
- Minnesota's legislature has passed a bill banning AI tools that create fake nude images, now heading to Governor Walz for signature
- Victims can sue the creators of these apps directly, not just the people who use them
- Sets a precedent for holding AI tool developers legally accountable for how their products are weaponized
The Signal
Most "AI safety" bills talk big and regulate nothing. Minnesota just passed legislation that actually has teeth. The bill doesn't stop at criminalizing deepfake revenge porn or non-consensual intimate images. It takes the fight upstream, to the developers building the apps that make this abuse trivial.
The key mechanism: victims can sue the creators of AI tools designed to generate fake nudity. Not just the person who used the tool against them. The people who built it, marketed it, and profited from it. This is liability at the source, and it changes the economics of building harmful AI.
"Minnesota's law doesn't just punish the crime. It defunds the pipeline."
This matters because the current landscape is flooded with apps explicitly marketed for creating non-consensual fake nudes. They're not hard to find. Many advertise openly. Some have millions of downloads. Until now, the legal response has been reactive: prosecute users after harm is done, usually under existing harassment or privacy laws that weren't written for AI-generated content.
Minnesota is trying something different. Make it financially and legally toxic to build these tools in the first place. The bill now heads to Governor Tim Walz, who is expected to sign it. If he does, Minnesota becomes the first U.S. state to ban not just the use, but the creation and distribution of AI apps purpose-built for generating fake intimate images.
The timing is significant. Deepfake porn has exploded in the past 18 months, driven by open-source image models and fine-tuned LoRAs that can generate realistic nudes from a single clothed photo. The barrier to entry is near zero. The harm is immediate and disproportionately targets women, minors, and anyone with a public online presence.
Key dynamics this bill targets:
- App developers who profit from subscription models for "undressing" apps
- Platforms that host or distribute these tools without meaningful content moderation
- The growing cottage industry of AI services marketed explicitly for harassment
The Implication
If other states follow Minnesota's model, we're looking at the end of the current free-for-all in harmful AI tools. Developers will need to weigh legal exposure against market demand. Some will geofence. Some will shut down. Some will move offshore and become harder to access, which is still a form of harm reduction.
For the broader agent economy, this is a signal: governments are starting to understand that liability needs to attach to builders, not just users. If you're developing AI tools, expect this logic to spread beyond deepfakes. The question isn't whether regulation is coming. It's whether you're building something that can survive scrutiny when it does.