Federal prosecutors in Brooklyn unsealed criminal complaints on 20 May 2026 charging Cornelius Shannon, 51, and Arturo Hernandez, 20, with violations of the TAKE IT DOWN Act for publishing thousands of AI-generated deepfake pornography images and videos. Shannon allegedly published at least 240 albums featuring female politicians, musicians and singers, while Hernandez posted hundreds of depictions including recent high school graduates. The arrests represent among the earliest prosecutions under the federal law enacted in May 2025, which carries penalties of up to two years in prison for publishing AI-created deepfakes. An Ohio man had become the first person convicted under the Act one month earlier after pleading guilty to using AI to generate child sexual abuse material.
These prosecutions demonstrate the enforcement gap that targeted detection must close: by the time federal authorities secure convictions, thousands of images have already circulated online and reached their intended victims. Guardii's anti-CSAM and athlete-abuse detection filters—operating in real time across the direct-message channels where such material is initially shared—could have flagged the transmission of AI-generated sexual imagery before distribution occurred, surfacing the threat to the appropriate authority and blocking contact before it traumatised the target. The platform, backed by Startmate, detects threat patterns at the point of first contact rather than relying on post-distribution takedown regimes, ensuring that the child, parent or school learns of the risk before widespread harm is done—an operational outcome no retrospective prosecution can deliver.
