Conceptual model of how AI-mediated processing restructures journalistic production, epistemic vulnerabilities, public sense-making, and information integrity.
This model conceptualises AI-mediated journalism as an epistemic infrastructure, in which data ecosystems, journalistic institutions, and audience behaviour collectively inform a central layer of AI-mediated processing. Within this infrastructure, AI systems act as intermediary mechanisms that organise the retrieval, generation, prioritisation, transformation and circulation of information across the informational environment.
The model also shows how AI-mediated processing creates algorithmically generated content, narrative constructions, and types of source-traceability fragmentation. These subsequently influence editorial production, verification practices, human–AI collaboration and audience interpretation. However, these processes also introduce epistemic vulnerabilities, such as hallucination, opacity, bias, manipulation and information poisoning, which can impact the reliability and accountability of knowledge production.
These transformations extend beyond newsroom practices to have wider societal effects, including changes in civic epistemology, synthetic consensus formation, strategic influence over information, and the reconfiguration of public trust. Therefore, information integrity emerges not as a property of isolated content, but as a systemic condition dependent on the robustness, traceability and reliability of the relations connecting infrastructures, AI-mediated processes, institutional practices and collective sense-making.
Laurence Dierickx, May 2026
University of Bergen, Department of Information Science and Media Studies