There's plenty of instances where the content produced by chatbots isn't factual and is also based off of content that somebody else worked to produce.
Well I don't know the full content of sparknotes. As far as I can remember, it provided summaries and analyses of literature. Those summaries obviously made reference to the original work. And those analyses were most likely not original & common enough to not need sourcing.
I'm not entirely sure what you're claiming. So because we can summarize books, we should allow the chatbot to regurgitate content and not make reference to the original?
It depends on the context, all copyrights are subject to fair use. You can use exact copies of content as a human, for example when educating, critiquing, or commenting on.
Why should machines built and run by humans be treated differently?