Interesting! I just tried it with the Twilight example. After a few more questions ChatGPT confessed that it didn't actually have access to the full text of Twilight, and agreed to warn me in the future if it was providing a quote for a text it didn't really know. That sounded good, so I immediately asked it to provide a quote from another Twilight chapter and it responded with "Certainly!" and the usual hallucinations.