Open this publication in new window or tab >>2025 (English)In: Telecommunications Policy, ISSN 0308-5961, E-ISSN 1879-3258, Vol. 49, no 5, article id 102952Article in journal (Refereed) Published
Abstract [en]
This study examines the role of news media in the context of generative AI-enhanced search engines, focusing on the 2024 Taiwan presidential election. Using Microsoft’s Copilot, we conducted a comparative analysis by prompting election news in five languages: English, Traditional Chinese, Simplified Chinese, German, and Swedish. While Copilot uses mainly professional news media, provides quick access to synthesized information, and exhibits source transparency, it frequently creates misinformation and misattributes news sources. The analysis highlights variations in Copilot’s sourcing behavior, showing a strong reliance on English-language sources, particularly those from the UK and US, across different prompting languages. Such reliance raises concerns about the homogenization of information and the marginalization of regional perspectives. The study underscores the critical role and dilemma of news media, which, while serving as authoritative sources in democratic societies, must navigate an increasing AI-mediated information ecosystem to maintain autonomy vis-à-vis powerful technological infrastructures. By evaluating Copilot’s sourcing practices and misinformation prevalence, this research contributes to the discourse on AI’s impact on news dissemination, media diversity, and democratic processes. Specifically, we discuss the consequences of two approaches available to news media to prevent their content from being used without compensation: opting out of crawling (“platform counterbalancing”) or establishing partnerships with AI companies. Current regulatory efforts, including copyright reforms and the EU AI Act, fall short of safeguarding journalism or regulating AI. We propose policy and regulatory recommendations to improve transparency, factual correctness, accuracy in source attribution, and accountability in AI-generated content, supporting informed citizenship in the digital age.
Place, publisher, year, edition, pages
Elsevier, 2025
Keywords
AI governance, Digital medium ecosystem, Generative search engine, Language model, Large language model, News media, News retrievals, Political news, Presidential election, Sourcing, Artificial intelligence
National Category
Media and Communications
Research subject
Media and Communication Studies
Identifiers
urn:nbn:se:kau:diva-104726 (URN)10.1016/j.telpol.2025.102952 (DOI)2-s2.0-105001927750 (Scopus ID)
Funder
Swedish Research Council, 2022-05392
2025-06-042025-06-042025-06-04Bibliographically approved