AI-generated child sexual abuse content is surging — and overwhelming law enforcement

AI-generated child sexual abuse content is surging

A wave of synthetic child sexual abuse content generated by artificial intelligence is overwhelming global child safety watchdogs and law enforcement agencies. In just the first half of 2025, the Internet Watch Foundation (IWF) has flagged 1,286 AI-generated child sexual abuse videos—compared to just two over the same period in 2024. Meanwhile, the U.S.-based National Center for Missing & Exploited Children (NCMEC) says it has already received 485,000 reports of AI-generated CSAM (child sexual abuse material) this year, a staggering increase from 67,000 in all of 2024, the New York Times reported.

“The technology has hit a tipping point,” said Derek Ray-Hill, interim CEO of the IWF. “There is an absolute tsunami we are seeing.”

Generative AI tools—once hampered by glitches like extra limbs or jerky video transitions—can now generate smooth, detailed, photorealistic images and videos that closely resemble actual child abuse. Analysts say newer content shows convincing backgrounds, realistic human proportions, and even complex scenes involving multiple individuals. In underground web forums, offenders are now praising the realism of this material and openly sharing tools to produce more.

Some of the new content splices real images of children scraped from school websites and social media with synthetic content. Others are entirely fictional, yet realistic enough to fool the human eye and evade moderation systems.

The Department of Justice says that criminal groups are using AI tools not only to generate imagery but also to blackmail children. Some apps allow users to create deepfake nudes of real people—including minors—who are then threatened or coerced into further acts. DOJ officials warn that these tactics are increasingly common, and that the availability of such tools is growing.

“There are apps that strip clothes off kids. Others generate entire libraries of fake abuse images,” said one DOJ official. “This isn’t future tech—it’s happening now.”

While U.S. federal law prohibits child sexual abuse content—real or synthetic—many state laws are still catching up to the technology. More than three dozen states have passed legislation targeting AI-generated child abuse content in recent years. But courts are only beginning to navigate how the First Amendment applies to fully synthetic content.

In one closely watched case, a Wisconsin man accused of producing 13,000 synthetic CSAM images succeeded in getting a possession charge dismissed. A federal judge ruled that so long as the content did not involve real children, its mere possession in the home might be protected by the First Amendment. Still, the man faces trial for distribution, especially since he tried to send the images to a minor on Instagram.

Tech companies offering AI tools are under pressure to prevent abuse. Amazon Web Services reported 380,000 incidents of AI-generated CSAM so far in 2025, while OpenAI reported 75,000 and Stability AI fewer than 30. All say they are improving their systems to detect and block this material.

Stability AI, whose image model Stable Diffusion came under fire last year for training on problematic datasets, said it has since implemented tighter safety filters and clearer oversight.

“We are deeply committed to preventing misuse of our technology,” the company said in a statement.

While AI-generated CSAM remains a small share of the total material flagged online—real abuse still accounts for the majority—analysts warn that the shift in quality and quantity could soon overwhelm investigators. Resources are being diverted to examine synthetic material, even as police continue chasing real-life predators.

“There is a finite capacity, and AI-generated material pulls time and manpower away from pursuing actual abusers,” Ray-Hill said. “The tech is accelerating, but the legal and enforcement frameworks are lagging dangerously behind.”