mother and daughter look at computer with shocked faces

The term “Artificial Intelligence” feels inescapable. From bulletin boards to YouTube ads, AI seems to be everywhere. It can be tempting to dismiss the technology as a passing fad and disconnect from the overwhelming number of conversations about it. Despite all the high-and-mighty “AI will revolutionize society” hype from big tech CEOs, AI harms real children today. One in 17 children is a victim of sexual deepfakes (Thorn 2025).

What are sexual deepfakes? 

Generative AI is advancing rapidly. It is a tool that can create original content such as images, videos, and text in response to a human’s request. This has enabled people to produce increasingly realistic “deepfakes.” According to the Alan Turing Institute, a “deepfake” refers to an “AI-generated video, image, or piece of audio content designed to mimic a real-life person or scene.” Deepfakes are typically created with the intent of deceiving an audience for various purposes. For example, someone may post a deepfake of their favorite musician singing a song they’ve never sung before. Or, they could post a photo of their friends dressed in ridiculous outfits. While deepfakes can be harmless, they can also be used for malicious and criminal purposes. The most harmful of these is to create non-consensual sexual imagery, including sexual deepfakes.

“Deepfake Nudes” or “sexual deepfakes” refer to the manufactured media creations that portray real people in sexually explicit or provocative activities. Using generative AI technology, perpetrators superimpose the faces of adults and/or children onto the bodies of people engaged in sexual conduct. Perpetrators can also now create non-consensual images from scratch. They also use “nudify” apps to edit existing pictures of adults and children to appear like they are engaging in sexual acts. Sexual deepfakes involving children can also be referred to as AI-generated CSAM (AIG-CSAM). In addition to sexual deepfakes, AI-generated CSAM also includes images that aren’t meant to resemble real, identifiable children. These images are made using AI tools trained on actual CSAM of children who have been exploited. 

Who Creates Deepfakes?

While adults are the primary perpetrators of sexual deepfakes, an alarming number of children are also participating in the abuse. According to a 2023 survey by the child sexual abuse prevention organization Thorn, 11% of respondents (aged 9-17) believed their friends or classmates engaged in deepfake creation. Schools face an increasing problem of students creating and spreading AIG-CSAM among their peers. For example, in the fall of 2025, a Louisiana middle school expelled a 13-year-old girl after she hit one of the boys who allegedly spread AIG-CSAM images of her. AIG-CSAM is on the rise, with one assessment showing a 6,345% increase in reporting from early 2024 to early 2025. Despite not depicting a real situation, deepfake nudes and AIG-CSAM are just as harmful and dangerous as revenge porn and real child sexual abuse material.

The Origin and Evolution of the Deepfake

The term “deepfake” (a portmanteau of “deep learning” and “fake”) was coined in 2017 on Reddit, when users swapped faces into sexually explicit videos and images. However, the concept of deepfaking can be traced back to the 1990s when people tried to create realistic human images with CGI. In 2014, computer scientist Ian Goodfellow and his team created the machine learning framework known as Generative Adversarial Network (GAN). This revolutionary technology ushered in a new era of artificial images, audio, and video by enabling machines to create media in previously unseen, realistic ways.

Public concern rose in the late 2010s and continued to increase as generative AI technology advanced in the early 2020s. The public launch of the text-to-image tech “Stable Diffusion” in 2022 changed the deepfake landscape. Now, anyone could create high-quality, realistic images using regular consumer hardware, not just specialized labs. In recent years, deepfakes have become more common. Since 2025, they are a major disruptive force in daily life.

xAI’s Facilitation of AI-Generated CSAM: 

Deepfakes have become more common because the technology has been democratized. Generating a deepfake requires fewer skills, resources, and technical know-how. This is due to less regulated AI models, open-source software, and affordable cloud services. For example, despite a pledge to stop the creation of abusive deepfakes, Elon Musk’s AI image generator Grok still allows users to create deepfake porn and AIG-CSAM. Musk’s company, xAI, faces an increasing number of lawsuits and global regulatory probes.

In July 2026, the state of Minnesota passed a law against technology companies facilitating “nudification” of images and videos through AI. xAI filed a lawsuit to block it, aruging for their right to continue creating easy access to nudification technology.

On this note, ENOUGH ABUSE is boycotting Musk’s Grok-integrated social media platform, X. We made a final post tagging both X and Grok to let them know we were leaving their platform due to their failure to protect children and their enabling of AIG-CSAM. 

But AIG-CSAM Isn’t “Real,” So Why Is It Still Harmful?
girl holding phone upset

While AIG-CSAM may not depict real situations or an identifiable child, children are still victimized and traumatized. The mass creation and spread of AIG-CSAM normalizes child sexual abuse and psychologically endangers children. The 13-year-old girl in Louisiana, whose images were shared by her classmates, described the teasing as relentless. She stated that because the images were shared on the social media app Snapchat (which automatically deletes messages seconds after they are viewed), the school principal doubted that they existed.

Furthermore, deepfakes have made it much easier for perpetrators to “sextort” victims. “Sextortion” refers to the crime of threatening to non-consensually distribute sexual or intimate images unless a perpetrator’s demands are met (i.e., sexual or financial). Traditional sextortion relied on tricking individuals into sending sexually explicit imagery of themselves and then demanding payment or sexual favors. However, deepfakes now allow perpetrators to generate hyper-realistic images, videos, or text that are both embarrassing and traumatizing.

According to an article published in early 2025, sextortion has been involved in at least 30 suicides of teen boys since 2021. As NCMEC explains, AIG-CSAM harms children in a multitude of ways, including harassment, bullying, and psychological harm. Although the imagery may not be real, the consequences of its proliferation and normalization are physical and tangible. 

Dealing With AIG-CSAM Is Difficult

One of the challenges of child abuse crimes committed with generative AI is that it is difficult to pinpoint exactly how the AI was used. For example, NCMEC states that “more than 133,000 reports have indicated a GAI [Generative AI] nexus but without enough information to know how the GAI was being used in the exploitation of a child.” There is a discrepancy between the rates at which companies and NCMEC identify and categorize generated content. While NCMEC categorized over 158,000 images and videos as AI-Generated from January 2023 to December 2025, electronic services annotated a little over 11,000 pieces of imagery. Labeling and categorizing these images is crucial to law enforcement’s efforts to prevent abuse. Thus, the companies’ slow pace endangers more children.

In addition, legal nuances restrict the routes prosecutors can take to enact justice. For example, if an image depicts a real child (AI-generated or not) it can be prosecuted under federal CSAM laws. However, if an image is entirely generated with AI (and does not depict a real child) it could instead be prosecuted under federal Obscenity laws. These statutes contain significantly different penalties, thus creating an inconsistent sentencing process, and subverting prosecutors’ accountability mechanisms. Also, many crimes of possession, creation and distribution of CSAM are prosecuted at the state level, which means state laws that fail to include AI-generated images and videos in their CSAM definitions also prevent prosecution.

Fighting Against AIG-CSAM

ENOUGH ABUSE® map showing 46 states that have criminalized AI-generated CSAM in blue.

Creating sexual deepfakes of children is never an invisible crime. The victims are often traumatized and retraumatized as these images and videos spread transnationally across platforms, forums, and the broader internet. Perpetrators who create and share sexual deepfakes of children must face the same legal punishments as perpetrators of traditional CSAM. As of August 2026, all but four states have adopted laws that criminalize AIG-CSAM and other technologically modified CSAM by including AI-generated images and videos in their CSAM definitions. These four states include Colorado, Massachusetts, Ohio, and Vermont, along with the capital city of Washington, D.C.

However, even some states that criminalize AIG-CSAM have deeply flawed laws. For example, Hawaii and several other states require that the AIG-CSAM include an “identifiable child” in order to prosecute a perpetrator. Similarly, some states have very detailed statutes, but their language does not include all computer-generated images. For example, a Colorado law uses the term “reproduced visual material,” which excludes AIG-CSAM that was “produced” or “created.” 

Pending Bills To Protect Children

Overall, sexual deepfakes of children and AIG-CSAM represent a new and volatile challenge in the fight to end child sexual abuse. It is important to move past preconceived notions that AIG-CSAM is not harmful because it is technically “fake.” Experts have depicted how AIG-CSAM endangers children, harms victims, and normalizes child sexual abuse. As AI technologies advance, we need to lobby for stronger policies that protect children across all levels of government. As such, please learn more about pending bills such as H.875 in Vermont, H.786 in Ohio, HB47 in Alaska, and S.2633 in Massachusetts. If you are from one of these states, help advocate for these bills! Tell lawmakers that you want to see sexual deepfakes of children and AI-generated CSAM outlawed. 

In addition, The Federal ENFORCE Act of 2025 seeks to close the legal loopholes mentioned in the section above regarding sentencing discrepancies. It would modernize federal statutes and strengthen sentencing requirements, ensuring that perpetrators face consistent and serious consequences. The ENFORCE ACT passed the Senate on December 16, 2025 and now needs to pass the House of Representatives. You can call your U.S. Senator and Representative and ask them to support this bill! 

Where to Learn More 

The fight to end child sexual abuse begins when all of us engage in the discussions of our time, including how to counter the ways the latest technology is used to harm children. To learn more about how to protect children online, visit our Online Child Sexual Abuse page, or read our related blogs:

About ENOUGH ABUSE:

ENOUGH ABUSE® is the nation’s oldest citizen-based child advocacy organization, with a 65-year history of improving the lives of vulnerable children. We work in states, nationally, and with international partners to ensure every child’s right to a childhood free from abuse and exploitation.

For over two decades, our primary focus has been to prevent the sexual abuse of children in their homes, communities and online through our effective prevention trainings, education and advocacy.