There is a very simple solution: Use Microsoft proprietary code for training the model. Keep your hands off open source code. There you have the "most responsi

There is a very simple solution: Use Microsoft proprietary code for training the model. Keep your hands off open source code. There you have the "most responsible way". The GPL should be updated to prohibit code to be used for "learning" (i.e., regurgitating copyrighted fragments).