Expecting Humans, Encountering Machines: An Expectancy Violations Theory Approach to Generative AI in Journalism

Drawing on Expectancy Violations Theory (EVT), we conducted 60 semi-structured interviews to examine (1) how news audiences in Australia and Germany form expectations about journalists’ use of generative AI (GenAI); (2) the antecedents shaping those expectations; and (3) how they evaluate perceived expectation violations. Participants envisioned journalism as a human-centered practice, rejecting autonomous GenAI use in core reporting functions. Transparency, disclosure, and oversight (whether regulatory or newsroom-led) were treated as conditions of legitimacy. Expectations were shaped by institutional trust, everyday experiences with AI, and broader reputations of AI technologies. Participants described negative violations linked to perceived ethical lapses, though some reported positive violations when AI enhanced accessibility or efficiency. Our findings demonstrate how audience responses to GenAI are not merely technological assessments, but relational evaluations grounded in normative expectations about journalism.