Extracting Training Data From Fine-Tuned Stable Diffusion Models
New research from the US presents a method to extract significant portions of training data from fine-tuned models. This could potentially provide legal evidence in cases where an artist’s style…