COTTONWOOD HEIGHTS — A Cottonwood Heights man was arrested Tuesday after police he may have thousands — or possibly millions — of images of child sex abuse material stored on his hard drives, and has also been creating lifelike AI images of child sex abuse material.

Russell Kenneth Gatrell, 45, was booked into the Salt Lake County Jail for investigation of five counts of aggravated sexual exploitation of a minor.

The investigation began on June 25, when the FBI’s Child Exploitation and Human Trafficking Task Force began investigating a person sharing child sexual abuse material.

“Investigators received 139 files from an IP address in Cottonwood Heights,” a police booking affidavit states. On Tuesday, investigators served a search warrant at Gatrell’s residence.

“Gatrell admitted to viewing, downloading, and sharing child sexual abuse material from his residence. Gatrell admitted to storing the child sexual abuse material on a network storage device which contained upwards of 15 hard drives totaling approximately 280TB of storage,” the affidavit states.

Gatrell told investigators he has an “unknown” amount of child sex abuse material stored on the hard drives, and estimated the total amount could be more than 100GB, according to the affidavit.

“It is believed that, at a minimum, Gatrell has been storing child sex abuse material across 280TB of storage, which could store approximately 56 million to 280 million photographs depending on the quality,” the affidavit says.

Gatrell has also been using AI to create child sex abuse material from stock images, according to investigators.

“Computer scientists with the FBI conducted an initial review of the AI Gatrell was using and determined it was incredibly advanced and that Gatrell was likely using images of identifiable children from the internet to create this child sex abuse material,” according to the affidavit.

The investigation into exactly how much child sex abuse material Gatrell has allegedly collected was ongoing as of Tuesday.

The Key Takeaways for this article were generated with the assistance of large language models and reviewed by our editorial team. The article, itself, is solely human-written.