I dont get it - has the copyright problem with AI generated code trained on other people's repositories been solved yet?