AI byline, human content: Exploring how source and message framing shape news perception

Drawing on Social Identity Theory, this study examines how people engage with artificial intelligence (AI)-generated news on a socially sensitive topic, such as child abuse, under different message framing conditions. We investigate how news source (human vs. AI writer) and message framing (emotional vs. factual) shape audience perceptions of identity threat and writer sincerity, as well as behavioral responses to the news and to AI more broadly. An online experiment with U.S. adults (N = 401) showed that while perceptions of identity threat did not differ across conditions, an AI writer was consistently perceived as less sincere than a human writer, particularly when emotionally framed. Lower perceived sincerity of the AI writer had downstream effects on attitudinal and behavioral responses, including reduced AI acceptance, lower willingness to pay for news, and increased AI aversion. By highlighting the interplay between source perceptions and message framing, this study offers novel insights into how concerns about the sincerity of information sources shape engagement with AI-generated content.