By admin 
PHOENIX (AZFamily) — The key evidence against William Powderly appears on line 36 of the Chandler police officer’s probable cause statement.
Read more Tucson charter school teachers vote to unionize
Powderly, 44, is the first person charged in Arizona with possessing AI-generated child sexual abuse material under a state law passed last year.
The officer’s affidavit says Powderly used an AI face-swapping application to put a photo of a 16-year-old girl he knew onto sexually explicit images of another young girl.
That “face swap” allegation is critical, according to legal experts. If Powderly had created child sexual abuse material that was entirely AI-generated – without the likeness of a real child – his defense may have had a strong First Amendment defense based on a Supreme Court precedent from 2002.
As AI technology advances, lawmakers and prosecutors across the country are confronting a complicated legal question: When does AI-generated child sexual abuse material cross the line from protected speech to criminal conduct?
Legal questions emerge over AI-generated content
Two Supreme Court cases frame the legal debate.
In 1982, the court found in New York v. Ferber that child pornography is not protected by the First Amendment because of the physical and mental harm to real children.
But a 2002 ruling complicates cases involving virtual or computer-generated images.
In Ashcroft v. Free Speech Coalition, the Supreme Court struck down a broad federal ban on virtual child pornography. The court said the government may ban material involving real children, and it may ban “computer morphing” that digitally alters images of real children into something sexual. But the court drew a line at images that are entirely computer-generated and do not involve real children.
The justices wrote that virtual child pornography “is not intrinsically related to the sexual abuse of children.”
That distinction could matter in the Powderly case. Because prosecutors allege Powderly used the face of a real 16-year-old girl, the case appears closer to the kind of “computer morphing” the Supreme Court said governments may regulate.
But other prosecutions have encountered roadblocks when trying to prosecute people over computer-generated sexual images of children.
In Wisconsin, prosecutors charged Steven Anderegg in 2024 with several crimes for allegedly using a program called Stable Diffusion to produce entirely AI-generated child sexual abuse images.
A federal judge later dismissed one charge tied to private possession of the images, finding the charge unconstitutional because the images did not depict real children. However, the ruling did not end the case. Other charges were allowed to move forward, including allegations that Anderegg distributed obscene material and transferred obscene material to a minor.
Read more What Benson recall could mean for controversial aluminum smelting plant
Gregg Leslie, executive director of the First Amendment Clinic at the ASU College of Law, spoke about the complexities on the latest episode of “Generation AI.”
“If they’re saying that really the interest is in protecting those children, and there is no children to protect, there is no child to protect when it’s AI,” Leslie said. “You can’t prosecute somebody for that if the harm that you meant to protect doesn’t exist.”
Technology could change the legal analysis
In the 2002 case, Justice Clarence Thomas argued that technological advances might one day change the First Amendment calculus if the technology started to impede legitimate child pornography prosecutions.
“If technological advances thwart prosecution of ‘unlawful speech,’ the Government may well have a compelling interest in barring or otherwise regulating some narrow category of ‘lawful speech’ in order to enforce effectively laws against pornography made through the abuse of real children,” he wrote.
As AI images get increasingly more convincing, Leslie said that the argument could lead to stronger state regulations.
“That is the next step in child pornography prosecution is to say that just having this market out there creates an appetite and actually leads to greater dangers for real children,” Leslie said. “That’s a stronger argument for saying we still have to regulate it, no matter what, that it isn’t just about protecting the victim in the image, but it’s about protecting the victim who comes next.”
But he said even then, there will still be room to make a strong First Amendment argument in defense.
RELATED: New federal law targets AI deepfake porn as schools grapple with growing problem
For prosecutors, the strongest cases may be those involving a real child’s image, identity or likeness — like an AI face swap, deepfake or digitally altered photo.
The weakest cases may involve wholly AI-generated images, privately possessed, with no identifiable child victim.
See a spelling or grammatical error in our story? Please click here to report it.
Do you have a photo or video of a breaking news story? Send it to us here with a brief description.
Read more PCC grads celebrate in Tucson, but face uneasy job market