OPINION:
Child predators are using artificial intelligence platforms to digitally “undress” real children and place them in sexually graphic images and videos.
One high-profile example is Grok, the image generator developed by Elon Musk’s company xAI. The city of Baltimore is taking xAI to court, alleging that the company created a system where a family’s ordinary photos “could be ingested by Grok and transformed into sexually degrading deepfakes without their knowledge or consent.”
Grok is not an outlier. Law enforcement has found predators using all kinds of AI tools — from smaller platforms to massive open-source models such as Stable Diffusion — to generate nude images of children.
Teenagers are creating AI-modified nude images of their classmates. Recent cases took place in New Jersey, Pennsylvania, Louisiana and Ohio.
In April, an Ohio man became the first person convicted under a new federal law criminalizing nonconsensual deepfakes. Investigators found he created more than 700 obscene images of children and adults.
The risk is not confined to any one city or platform. It is no longer safe to post any photo or video of children online.
With a few lines of prompting and the click of a button, predators can now download parents’ perfectly innocent, fully clothed photos of their children playing sports or celebrating a birthday and instantly, anonymously turn them into abusive, sexualized content.
That will not change until Congress steps in to hold AI companies and social media platforms accountable.
Online child sexual abuse had already reached unprecedented levels before the widespread adoption of generative AI. In 2024, the National Center for Missing and Exploited Children received roughly 50,000 reports of suspected child sexual abuse material every day, most depicting prepubescent children.
Reports of sextortion, in which predators threaten to release explicit material of a minor, similarly surged.
Artificial intelligence has made it impossible for law enforcement to keep up with the scale and speed at which predators can now create and distribute child sexual abuse material. British regulators, for instance, are investigating thousands of sexualized images of women and children that Grok generated on X.
Technology companies often insist they can police this behavior themselves. The evidence suggests otherwise. For years, leading platforms such as Meta and Google have pledged to prioritize child safety while resisting binding safeguards.
Stanford University researchers showed that large, web-scraped datasets used in AI development have contained child sexual abuse material that filtering systems failed to catch.
One dataset used to train the wildly popular AI image generator Stable Diffusion contained thousands of suspected child sexual abuse material images.
No federal standard requires AI developers to monitor their platforms for child sexual abuse material and immediately report it.
As a result, efforts to stop the creation of this content are largely reactive. Platforms typically act only after end users spot and report abuse — removing content once it has already been created, shared, copied and archived.
Companies are releasing increasingly powerful AI tools faster than they can meaningfully test, monitor or enforce protections. Grok was released in December 2024 and embedded directly into X, a platform with hundreds of millions of users.
Reports suggest it has likely generated millions of images by now.
With private safeguards failing, responsibility shifts to lawmakers. Congress needs to act decisively. At a minimum, lawmakers must establish clear, enforceable rules that prohibit AI tools from creating sexualized images of children.
We cannot rely on the goodness of the people using AI tools as a safety strategy. Some users will inevitably employ them to create abusive images. Congress must ensure that the companies building these tools put defenses in place that make it impossible to create sexualized images of children using the technology and hold them accountable if they fail to do so.
Consistent legal consequences would force companies to design child protection safeguards from the start, not as an afterthought.
Every day that lawmakers delay is another day predators spend creating more abusive material. Children should not have to sacrifice their privacy or innocence because lawmakers failed to keep pace with technological advances.
The question before Congress is not whether regulation is a perfect solution, but whether inaction is an acceptable response. When it comes to protecting children, it is not.
• Teresa Huizar is the CEO of Washington-based National Children’s Alliance, the nation’s network of nearly 1,000 children’s advocacy centers, providing justice and healing through services to child victims of abuse and their families.
Please read our comment policy before commenting.