In the end, it’s important for those who make lying machines to be held accountable for the falsehoods they spread. Not to mention the copyright infringement. If I write code and it uses it for training, if it’s licensed under the GPL, all derivative code is copyleft. Am I wrong there? I know Microsoft trains Github Copilot on public code, does that mean they’re respecting the license of the code they’re using for training? I have so many questions.