There’s now an iPhone app to do Stable Diffusion AI image generation. Free, runs locally, no data collected. When you first open it, it will download about 2 GB of model data. It’s for iPhone, but works well on my ipad air 5 (M1) and takes a little under a minute for 512x512 pixels. Lots of fun, though I need to learn how to talk to it. Simple phrases don’t even approach the uncanny valley (but I kind of like the mouse with toes on the tail tip…)
Deviant Art are pressing for industry adoption of their opt-out tags embedded in their user pages which would preclude any scraping of images for datasets.