by Logan Finney, Idaho Reports

The Idaho House of Representatives approved legislation on Tuesday that aims to crack down on images of child sexual abuse that have been generated through the use of artificial intelligence.

Republican Reps. Bruce Skaug of Nampa and Dori Healey of Boise jointly sponsored House Bill 465. It would add a new section of code to state law regarding visual representations of child sexual abuse created using generative AI or machine learning.

“The question I’ve been asked over and over again: Does it really matter if it’s not a real image?” Healey said. “It continues to satisfy the sexual cravings of lewd adults. The images are progressing so rapidly that, when this first began, it was really simple to tell the difference between an artificial image and a real image. Today, prosecutors are having a difficult time, our people that investigate this are having a difficult time, making it more and more difficult to get this off the street.”

Machine learning systems for image generation are typically trained on massive datasets of real images. In December, researchers at Stanford University identified over a thousand verified instances of child sexual abuse material, or CSAM, in a public open-source dataset used to train some popular AI text-to-image generation models.

Skaug said he interviewed investigators from the Internet Crimes Against Children taskforce at the Idaho Attorney General’s Office while working on the bill.

“You cannot tell, as a prosecuting attorney or an investigator. Is this a real child or is this AI? It has the same effect.” Skaug said. “And the defense in those cases is: Prove to me this isn’t an AI child. If you’re defending that person, they get off of the charges.”

HB 465 passed the House unanimously and now advances to the Senate.

“We’re trying to catch up to the evil side of AI,” Skaug said. “Artificial intelligence can be a wonderful thing, but it’s like any tool. It can be misused, and this is being misused to hurt children.”

---