The authors found that the LLM-rephrased tweets tend to display a more neutral sentiment than the original texts, a blunting effect that occurred irrespective of the prompts employed or the ...