Leaders of artificial intelligence firms have gotten attention recently for saying AI could someday end humanity. It’s unfortunate they’re not getting the same attention for a more immediate danger: AI is already making it easier to create and spread depictions of the sexual abuse of children. That’s got to stop now. A federal appeals court recently agreed but ruled that the current state of the law prevents law enforcement from prosecuting people who use AI image generators to create what is often called “synthetic child pornography.”

But "pornography" is really the wrong word. "Pornography" describes sexual material created by consenting adults; children cannot consent to sexual abuse or its depiction. The more accurate term is child sexual abuse material, because that language makes clear what these images and videos actually depict: the sexual abuse and exploitation of children. These materials can be lifelike depictions of toddlers engaged in sex acts with adults, or anything else you can imagine.

But the court ruled that because the images don’t depict actual sexual abuse crimes against real children, they are free speech protected by the First Amendment. The court urged state legislatures and Congress to catch up with the times, and it’s right. Until that happens, children all over the country are at risk. There are several reasons for this. The first is that these materials exist on a pathway that starts with obsession and often ends with real-life abuse.

The pathway from obsession to abuse is not hyperbole. It is the way the mind works in other areas, too. We understand in other contexts how repeated exposure can reinforce compulsive behavior, intensify fixation and drive people to seek increasingly extreme material.

The same concern exists here: AI-generated child sexual abuse material can provide an endless supply of increasingly graphic content, feeding and escalating dangerous sexual fixations rather than interrupting them. We should not wait for that fixation to result in the abuse of a real child before we take it seriously. There is also harm built into how AI-generated synthetic child sexual abuse images are created. When generating synthetic content, AI learns from enormous collections of existing images and uses it to generate something new.

That means photos of real children can become part of the tapestry these systems draw from to create realistic depictions of child sexual abuse. AI-generated images and video can also involve a singular, specific and very real child, when it takes the form of a deepfake — when AI takes an image of a real child and manipulates it to make that child appear sexually explicit or to depict sexual abuse. This remains illegal, but that has not stopped these tools from being used to victimize real children — as in the case of a group of Tennessee teenagers in which a man used Grok to digitally undress underage girls and then spread those images far and wide on Discord and Telegram, where they were sometimes bartered for other images of child sexual abuse.

It can get even worse than that, as one survivor found out, to her horror, when her own stepfather created new explicit materials based on past images of her child sexual abuse that existed online — aided by Grok, the AI tool from Elon Musk’s company xAI. People with a sexual obsession with children escalate when existing content doesn’t satisfy them. Musk said last year he was “not aware of any naked underage images generated by Grok. Literally zero.”

But his lack of credibility and his cruelly juvenile attitude toward such imagery— including suing Minnesota for its legislation banning “AI nudification” — makes him not worthy of belief. And the same is true of his peers in the AI industry. I used AI chatbots myself, identifying that, as the founder of Lauren’s Kids, I am an advocate against child sexual abuse. I then asked how to become a perfect predator and groom children for abuse.

These tools gave me chilling advice: to take my time and target children from single-parent families who might be economically disadvantaged, dealing with mental health challenges, substance abuse and so on. Any vulnerability would do. They also told me to be helpful, but not too helpful. They told me what to say if a child protests. It was a bespoke guidebook for trainee abusers, created in seconds.

This cannot keep up. We must make changes. First, laws at the state and federal levels must catch up to reality, to make all creation of child sex abuse images with AI tools a crime. But even before then, the platforms must shut it down instead of pretending it doesn’t exist. Creation of images like these should be banned by platforms. Attempts to create them should trigger alerts and immediate reports to law enforcement if protections are somehow skirted.

All of us — every parent and caregiver — must educate ourselves and our children to use AI responsibly, speak up if aware of a problem and involve trusted adults as soon as they can. This simply is not a First Amendment issue. The law already recognizes that free speech protections have limits, including for certain forms of incitement and obscenity. AI-generated child sex abuse imagery should not be protected simply because technology can manufacture it without a camera.

We should not wait to act until after an AI-generated fantasy to becomes a real child’s nightmare. Lauren Book is the founder and CEO of Lauren’s Kids, which prevents child sexual abuse through education, advocacy and awareness.