Post by @alexouyouyou • Hey

Foundation models are trained on copyrighted data and have been shown to regurgitate verbatim copyrighted material. But whether a generated output infringe