A federal court ruling could challenge the constitutionality of part of the Wisconsin state law that bans AI-generated child pornography.
Late last month, a three-judge panel ruled that the First Amendment protects a person’s right to privately possess images or videos of child sexual abuse, if the material does not depict a real person and remains in the home.
The appeals court ruled in favor of a La Crosse County man, who was criminally charged with using an open- children
Understanding Wisconsin, Together.
WPR’s “Wisconsin Today” newsletter keeps you connected to the state you love without feeling overwhelmed. No paywall. No agenda. No corporate filter.
The ruling only applied to his criminal charge for possessing the materials, but he still faces prosecution for producing and distributing them.
Wisconsin’s 2024 law created a new crime in the state for receiving, distributing, producing or possessing obscene material that depicts a purported child engaging in sexually explicit conduct.
University of Wisconsin Law School Professor Ben Sobel said the state law remains in place and enforceable, but this new federal court ruling created a legal precedent that defense attorneys could use to challenge the aspect of the law only pertaining to possession.
He joined WPR’s “Wisconsin Today” to explain the legal reasoning behind the ruling and what it means for regulating AI-generated child sexual abuse materials.
The following interview was edited for clarity and brevity.
Rob Ferrett: What did the court decide in this case, and what was its reasoning?
Ben Sobel: So it’s important to emphasize that we’re dealing with a narrow ruling. It’s by no means a get-out-of-jail-free card for the defendant. The ruling is that the First Amendment prohibits the prosecution of or the criminalization of simple possession of obscene material in the home.
We’re not talking about media that records an actual offense against children that took place in real life against an actual identifiable child. That material is not protected by the First Amendment. The possession of that material can be criminalized. Instead, we’re dealing with purely AI-generated virtual child sexual abuse material, and the rationale here is that it’s too intrusive under the First Amendment to reach into the home and regulate what media a person may possess there. But that rationale does not extend to regulating what kind of material someone distributes or produces.
RF: Does this decision jeopardize Wisconsin’s state law banning AI-generated child pornography?
BS: It’s really worth emphasizing again the narrowness of the present decision really only concerning private possession of this material. So only insofar as state legislation prohibits that is there a constitutional problem under the reasoning of this case. There remain many avenues for addressing this material that are outside the reasoning of this case.
I don’t think that this court decision is an obstacle to enforcing the law in the Wisconsin books, except insofar as it pertains to private possession in the home.
RF: How does production of these materials fall into a different category than possession?
BS: The judge declined to say that this First Amendment protection extended to production. The tension is, of course, if you are free under the First Amendment to possess obscene media, then presumably, you couldn’t say that it’s illegal to draw an obscene drawing in the privacy of your home. Yet, I think understanding that we’re dealing with a different technology and a really limited precedent from the Supreme Court, the district judge in this case declined to extend it to cover production.
RF: Is there a legal difference when the AI-generated child sex abuse material depicts a real, identifiable person as opposed to a generic image of a child?
BS: Yes, that’s absolutely a different issue here. The appellate court in this case noted that the government conceded that we’re not dealing with something that depicts an identifiable child or is even linked to one. Now I would question whether that’s actually true, in the sense that I’m sure that there are images of real children in the data that trained this model, but nevertheless, that’s the sort of facts that we’re accepting for the purposes of this case. That says nothing about depictions of particular identifiable individuals, whether they’re adults or children.
There are a few different concerns here that are worth disentangling. One is the concern that the proliferation of this hyperrealistic media will overrun existing infrastructure for identifying and reporting child sexual abuse like tip lines and stuff. It does seem like that infrastructure is really strained under this flood of new imagery. A separate question is whether this kind of imagery makes it more difficult to to prove cases against defendants for possession of actual depictions of child sexual abuse.
RF: Is the company that created the AI that people use to generate child sex abuse materials liable in any way?
BS: This is a notable case because it involved a program called Stable Diffusion, which is an openrol how it is used. Companies that have tighter control over their applications may impose more safeguards that prevent generating this kind of media. In contrast, something like Stable Diffusion is harder to regulate. It’s harder to control the sorts of stuff that it produces or that people can tweak it to produce
RF: Is there a way to regulate openuce this kind of material?
BS: I think that that remains a more difficult problem, particularly if you’re dealing with people that are outside of the jurisdiction of the United States. And so I think the answer is that we would face a similar problem with the enforcement of child sexual abuse laws more generally, which is that there may be wrongdoers outside of the reach of U.S. law enforcement.
RF: How should elected officials and policy makers be thinking about this issue?
BS: The arguments in favor of regulating this material because it leads to downstream harms, like facilitating the grooming of children or encouraging people to commit sexual abuse crimes, those are different from the arguements for regulating the media itself because it is obscene. And so, I think when drafting legislation or considering policy, it is always helpful to disentangle those distinct sorts of claims, which may receive distinct legal treatment and are more empirical questions versus moral and social questions.
In this case, there was a charge of distribution of obscene media to a minor, and so if the concern is that children in particular shouldn’t have access to this material, there is a law for addressing that. And of course, that’s distinct from a law that prohibits virtual child sexual abuse material solely because it is obscene.
