These videos started with the idea that working with diffusion models means you are building on a training data set of billions of images and clips and what if some of those training data “broke into” the resulting video. I wanted to lean into the misunderstanding that all these systems do is cut up previous works and jumble them into something that is solely derivative. The porous boundary between influence and extraction, always a gray zone in art, here becomes visibly literal. When using AI in your craft, are you creating a sovereign artwork or a statistical average of culture? Same as others