Abstract
<jats:p>Audiences increasingly encounter the news not through newsrooms but through chatbots, AI-generated summaries, and platform intermediaries that reproduce the form of journalism while omitting the human judgment that produces it. This paper asks what is lost in that substitution. We argue that a central but invisible casualty is epistemic labor: the contextual, accountable human work that converts plausible-sounding information into locally trustworthy knowledge. We examine this from an unusual vantage point, inside a local television newsroom in the American Midwest that is integrating an AI-assisted editorial tool, where the same class of technology audiences meet outside the newsroom becomes briefly visible at the human-AI handoff. Drawing on an in-progress, IRB-approved qualitative case study (embedded observation, workflow mapping, journalist interviews, and participatory co-design), we introduce the provenance gap: the systematic invisibility of epistemic labor in AI-mediated news, which leaves audiences unable to distinguish accountable local reporting from outputs that merely resemble it. We discuss implications for provenance signaling, local news sustainability, and the design of AI news agents.</jats:p>