It might, possibly, be a viable use case if the LLM produced the summary for an editor, who then confirmed it’s veracity and appropriateness to the article and posted it themselves.