3rd Grade Teacher Accused of using Yearbook Photos to create Explicit Images with AI

I’ve found an interesting story which highlights a very real issue that I feel needs to be discussed, but I also want to use it to emphasize how important it is that laws targeting synthetic/virtual child pornography are limited to materials depicting/made from real children.

If you use AI technology to create images of fictional children who do not exist, then you are fine, and have not impacted anyone, nor have you stimulated the market for actual CSAM.

If you use AI to take the photos/likenesses of actual, living children and superimpose them into sexual situations, then you are participating in the sexualization and sexual exploitation of actual minors by misusing their likeness in a way they cannot consent to.

Here’s the story.

Possession of Child Pornography, New Port Richey

Pasco Sheriff’s Office arrested an individual employed by Beacon Christian Academy, located on Star Trail in New Port Richey today. Steven Houser, 67, was arrested for possession of child pornography.

After receiving a tip, the Pasco Sheriff’s Office began an investigation which determined that Houser, a third grade science teacher at the school, possessed two photos and three videos featuring child pornography. The child pornography did not feature any students at the school.

In addition, Houser also possessed child erotica generated by an AI computer program which Houser admitted he used the yearbook photos of three students to generate. Houser resides in New Port Richey.

Keep it fictional.


I also want to iterate that AI-generated child pornography has the potential to completely eradicate the market for actual CSAM, though, we must ensure that such technology is used in ways that cannot be used to harm other people, and actual, earnest, good-faith attempts must be made to ensure that it is not done for this reason.

Criminals want to use it to exploit actual children, but far more sinister forces want to use this minority of criminals as a talking point to further disenfranchise a class of people and censor thoughts and free speech.

Don’t give these people ammunition.


While this might well be an inappropriate invasion of privacy (invasions of privacy can even be crimes), conflating possession of this artificial imagery with distribution is sensationalistic and misleading.

You are falling into the same trap a few pundits fell into. They focus on content. Talking about fighting content with War on Drugs like rhetoric (and they conflated the real and fictional). When you look at examples, it’s clear it isn’t about content, it’s about conduct. If they’re harassing someone, that is conduct (and we can arrest them for that). You are technically correct, someone should not try to make it look like an actual person. Nonetheless, you are largely missing the point.

Status quo. It’s OK to tease but not OK to please…