There are interesting signing processes in software being done for photography in camera to help combat this, ironically some utilise machine learning as part of the process. I wonder if something similar could be done for digital art