Post #4230176

IMO, the simplest solution to the ethics of training #AI on public web content is to always release the model open weights. You got training data from the public, the public deserves to own the result.