@hexbatch@OhOkKay@kenthompson@bookstodon@pluralistic I definitely agree that that is a good practice, either that or some other way of preserving drafts, notes, or evidence of incremental progress. (There isn't really anything special about Git in that regard; a commit history can easily be faked.)
But I also think it's not reasonable to expect, in general, that writers must have done this. There are, and (probably) always will be, many cases where an author doesn't have records of their incremental progress, and that can't mean they have no defense against an accusation of AI-powered plagiarism.