Reading code and learning from it doesn't violate copyright. Ensuring that copyleft licenses aren't broken must be handled the same way as with human developers.
I'm now guessing that you believe that LLM-training means storing the training data in the model. Am I correct?
[@simon_brooke](https://mastodon.scot/@simon_brooke ) [@nina_kali_nina](https://tech.lgbt/@nina_kali_nina )
