A federal appeals court has ruled that the First Amendment protects the private possession of AI-generated child sexual abuse material (CSAM), provided that the images do not depict real children, a decision that exposes a startling gap between decades-old Supreme Court precedent and modern artificial intelligence.
The 7th U.S. Circuit Court of Appeals ruled Tuesday that the government cannot prosecute Steven Anderegg for possessing AI-generated images depicting children engaged in sexually explicit conduct because Supreme Court precedent protects the private possession of such material.
Two of the three judges on the panel, in a separate opinion, acknowledged the disturbing implications of a ruling they felt bound by precedent to reach.
“We now live in an age where GenAI models can render images depicting the abuse of virtual children that are virtually indistinguishable from those depicting the abuse of actual children,” Judge John Lee wrote in a concurrence joined by Judge Joshua Kolar. “Given the relentless advancement in artificial intelligence models, we have some concerns about the lines these cases draw, but we are not free to redraw them ourselves.”
The ruling stems from the prosecution of Anderegg, a Wisconsin man accused of using an AI image generator to create thousands of sexually explicit images depicting children. According to prosecutors, an investigation began after Meta, the parent company of Instagram, reported to the National Center for Missing and Exploited Children that an Instagram user had sent apparent AI-generated CSAM to an account belonging to a minor. Anderegg was ultimately charged with producing and distributing the material, transferring it to a minor under 16, and possessing it.
A federal district court dismissed the possession charge, concluding that prosecuting Anderegg for privately possessing the AI-generated images violated the First Amendment. The Justice Department appealed that decision to the 7th Circuit. The appeals court affirmed the dismissal.
Crucially, the decision does not mean that everything Anderegg allegedly did is protected by the First Amendment. The charges involving the production and distribution of the material, as well as allegedly sending it to a 15-year-old boy, remain pending.
Still, the court’s decision emphasizes a legal distinction: If a person creates or obtains AI-generated CSAM through conduct that is otherwise prosecutable, the government is able to prosecute that conduct, but once purely virtual material is privately possessed inside the home, existing Supreme Court precedent prevents the government from criminalizing the possession itself.
The foundation for the conundrum dates back to 1969, when the Supreme Court ruled in Stanley v. Georgia that the government generally could not criminalize an individual’s private possession of obscene material inside his home.
Building on that foundation three decades later, the Supreme Court confronted virtual child pornography in Ashcroft v. Free Speech Coalition. In the 2002 decision, the justices struck down portions of a federal law that prohibited sexually explicit depictions of fictional children when no actual child was involved in their production.
The Supreme Court reasoned that the government’s strongest justification for prohibiting child pornography — protecting children from being exploited in its production — did not apply when no real child was involved. Those rulings were written long before today’s generative AI systems could produce photorealistic images with a few prompts.
The 7th Circuit said that distinction has become increasingly difficult to maintain. “Free Speech Coalition” was decided nearly 25 years ago, Lee wrote, and “the image-generation technology available today was likely unimaginable back then.”
The court also pointed to emerging evidence that AI-generated CSAM may not be entirely disconnected from the exploitation of real children. Lee noted that researchers have found CSAM in datasets used to train image-generation models, raising the possibility that some supposedly “virtual” material could ultimately be connected to real-world abuse. That issue, however, was not fully developed by the government in Anderegg’s appeal, leaving the court unable to resolve it.
The 7th Circuit’s opinion also revived a warning made by Supreme Court Justice Clarence Thomas more than two decades ago. In his 2002 concurrence in Free Speech Coalition, Thomas warned that technological advances could eventually make it difficult to enforce laws against child pornography involving actual victims if virtual images became sufficiently realistic.
Thomas wrote that if technological advances prevented prosecutors from effectively enforcing laws against illegal material, the government might eventually have a compelling interest in regulating a narrow category of otherwise lawful speech. “According to some experts, that day may have arrived,” Lee concluded, citing the growing volume of AI-generated material reported to the CyberTipline and the difficulties it presents for law enforcement.
The judges ultimately concluded that it is not their role to overrule or rewrite Supreme Court precedent. “The stakes are high,” Lee wrote, “and we must be mindful of the ever-accelerating advancements in image generation technology.” The decision therefore leaves the issue squarely in the Supreme Court’s hands.
.png)
.png)
