Beyond the Byline: Audience Expectations for AI Disclosure in News Media

As generative AI becomes increasingly common in news media, newsrooms are trying to find the most effective and reliable ways to communicate this AI use to audiences. This paper aims to unravel audience perspectives regarding the purpose of AI transparency, motivations for desiring transparency, expectations of journalists using AI, and perceptions of different AI labeling approaches to guide the implementation of effective and responsible AI disclosure. Through in-depth interviews with a sample of news consumers in the U.S (N = 20), we find that audiences are less concerned with the technicalities of generative AI content creation but rather, based on their expectations of journalists, they value labeling that signals human-AI collaboration and visible human oversight. By focusing on audience expectations and perceptions, we contribute practical and theoretical insights to the literature on AI labeling.