Í didn't realize Stable Diffusion has a folder where it saves all outputs you generate

So Í've been running SD via the Automatic1111 WebUI on my research institute's HPC cluster. If the state of Texas is going to spend millions of taxpayer dollars on creating a massive system for ML for bioinformatics and computational biology, Í was intending to make good use of it. The lure of 320GB of GPU on an individually dedicated node was too much, and soon enough Í was using it to produce loads of high-quality fetish wank-bait.

That all has come crashing down like the Hindenburg. Í thought Í was smart, saving the outputs that were good to a personal device. What Í neglected to realize was that this program automatically saves all outputs to another folder with my name on it, one that's sitting in my research lab's shared directory. Some busybody noticed that it was taking up a disproportionate amount of the lab's allocated cloud storage.

They opened it up and found around 8TB of high quality and absurdly degenerate smut. In addition, there's numerous Dreamcloud models and Loras just sitting there, each directly named for the porn star they were trained on, as well as ones designed to reinforce learning for the appearance of specific fetishes.

As of this afternoon, Í have been informed that access to all cloud and HPC services have been suspended, and Í need to be watching Outlook for an in-person meetings with PI, department chair, head of IT, and HR that will be scheduled for Monday.

Í'm a great employee otherwise, and have contributed to multiple papers that have been published in the top echelon of journals in the past year. Í have a flawless output record. Most importantly, Í never messed around with SD when working, so it's not like Í committed wage theft or anything.

Anyway, what's the verdict?

trans lives matter