I suspect it's because AI is specifically trained to be good at summarizing stuff, but the easiest way to check if it summarized something accurately is if the

I suspect it's because AI is specifically trained to be good at summarizing stuff, but the easiest way to check if it summarized something accurately is if the summary content matches/contains one or more specific claims from the source(s). With such a focus on accuracy and avoiding hallucination, they may have overfit on "repeat things you find verbatim when asked to summarize".