A deputy attorney general in New Jersey has been arrested on federal charges of producing and distributing AI-generated child sexual abuse material, according to charging documents unsealed this week.

The defendant, who worked in the state’s Division of Criminal Justice, allegedly used the Discord username “thegentlemanpervert” to share synthetic CSAM created using artificial intelligence tools. The case began when another Discord user flagged the material to the platform, which then reported it to the National Center for Missing & Exploited Children as required by federal law.

This isn’t a case of someone stumbling into legal trouble through ignorance. The defendant is a lawyer who prosecuted crimes. He would have understood exactly what he was doing and what the law says about it.

What the law says about AI-generated CSAM

Federal law prohibits creating, possessing, or distributing any visual depiction of a minor engaging in sexually explicit conduct, whether the image depicts a real child or not. The key statute, 18 U.S.C. § 2252A, covers both photographic images and computer-generated imagery.

Courts have consistently upheld these prohibitions. The Supreme Court’s 2002 decision in Ashcroft v. Free Speech Coalition did strike down part of the Child Pornography Prevention Act that banned “virtual” child pornography too broadly. But Congress responded by passing the PROTECT Act of 2003, which more narrowly targeted obscene material that appears to depict minors. That law has survived legal challenges.

The distinction matters: you can’t defend AI-generated CSAM by arguing that no actual child was harmed in its creation. The material is illegal regardless of how it was made.

Why these cases are increasing

Prosecutors and child safety advocates have warned for months that generative AI tools are making it easier to create realistic abuse imagery at scale. Some of these tools are purpose-built for creating explicit content. Others are mainstream image generators that bad actors have jailbroken or fine-tuned to bypass safety filters.

The technology has created two related problems. First, it’s harder for investigators to determine whether an image depicts a real victim who needs to be identified and rescued. Second, the volume of synthetic material flooding reporting systems threatens to overwhelm the infrastructure designed to catch actual abuse.

NCMEC reported a 300% increase in AI-generated CSAM reports in 2025 compared to 2024. The Internet Watch Foundation, which tracks online child sexual abuse material globally, has documented similar trends.

The Discord connection

Discord has become a recurring setting for these cases. The platform’s server structure, which allows users to create private communities with minimal oversight, has made it popular for distributing illegal content. Discord says it uses a combination of automated detection and user reports to identify and remove CSAM, and that it reports all instances to NCMEC as required by law.

But the volume problem is real. A single user can generate hundreds of images in a day using modern AI tools. Distribution happens quickly in private servers. By the time someone flags the content, it’s often been shared widely.

In this case, the user report system worked as designed. Someone saw the material, reported it, and law enforcement investigated. But that’s reactive, not preventive. The images were already created and distributed before anyone intervened.

What happens next

The defendant faces up to 20 years in prison on the production charge and up to 10 years on the distribution charge if convicted. He’s been placed on administrative leave from his position with the state.

The case also raises uncomfortable questions about screening and oversight. How does someone in a position of public trust, with access to sensitive criminal investigations, end up creating and sharing child abuse material? What mechanisms failed to catch this earlier?

More broadly, this case is going to fuel ongoing debates about AI regulation. Some advocates have called for mandatory safety measures in image generation models to prevent them from creating CSAM. Others want stronger requirements for platforms to proactively detect and report synthetic abuse material, not just wait for user reports.

The technology is evolving faster than the legal and policy responses. We now have widely available tools that can create realistic abuse imagery of children who don’t exist. The laws prohibiting that material are clear. But enforcement is struggling to keep pace with the scale of the problem.