A Wisconsin man facing charges for creating thousands of AI-generated images depicting child sexual abuse just caught a major legal break that could have chilling implications for how courts handle this emerging technology.
The Seventh Circuit Court of Appeals ruled last week that Steven Anderegg, 42, cannot be charged with possession of the disturbing material because the First Amendment protects his right to possess “obscene material” in his own home. The ruling, based on existing Supreme Court precedent, effectively creates a legal pathway for anyone to possess AI-generated child pornography as long as they keep it private.
Anderegg, a software engineer with expertise in AI models, first came to the attention of federal authorities in October 2023 when Meta flagged his account for transmitting apparent child pornography to another user on Instagram. The company’s monitoring systems, designed to protect children from predators on its platforms, detected the criminal activity and alerted the feds.
Investigators quickly obtained a search warrant and seized multiple cellphones and a laptop from Anderegg’s residence. What they found was disturbing: the software engineer had been using an AI model called Stable Diffusion, modified with third-party software, to generate thousands of explicit images depicting minors. The model would not have produced such content on its own. Anderegg had deliberately engineered it to do so.
Federal prosecutors charged him with producing, distributing, and possessing child sexual abuse material. The case seemed straightforward at the time.
But the appeals court’s decision changes that calculus. The judges based their ruling on two Supreme Court precedents. The first establishes the home as a protected space where individuals can possess obscene material without government interference. “If the First Amendment means anything,” the court explained in its ruling, “it means that a State has no business telling a man, sitting alone in his own house, what books he may read or what films he may watch.”
The second precedent dates to 2002, when the Supreme Court ruled that “virtual child pornography” where no actual child was involved could be legally possessed. At that time, computer graphics were primitive by today’s standards. AI-generated images didn’t exist. The technology has advanced dramatically, but the legal framework hasn’t kept pace.
Several judges on the panel expressed discomfort with the outcome their own ruling required. They indicated they want the Supreme Court to revisit its precedent on virtual child pornography given the new technological reality. But bound by existing case law, they had no choice but to rule as they did.
Anderegg isn’t walking free. He still faces charges for distributing the material and for manufacturing it. The production charge recognizes that he actively used his computer and technical skills to create the images. The distribution charge stems from the transmission that originally caught Meta’s attention.
But the possession charge is gone. Under this ruling, Americans can legally possess AI-generated child sexual abuse material in the privacy of their homes. They simply cannot share it.
The contradiction is glaring. Material deemed too horrific to transmit to another person is somehow acceptable to possess privately. The logic strains credulity and leaves a gap in the law that bad actors will surely exploit as AI technology becomes more accessible and more powerful.
The case highlights a growing tension between decades-old legal frameworks and rapidly evolving technology. What the Supreme Court envisioned in 2002 bears little resemblance to what AI can produce today. Until the high court revisits the issue, this ruling stands as the law in the Seventh Circuit.
