The Impact of LLMs on News Reception and Agenda Setting

This study examines the impact of large language models (LLMs) on news reception and agenda setting. Using a sequential mixed-methods design, three phases were implemented: LLM-assisted content analysis of 12,847journalistic pieces (2024–2026), focus groups with 94 participants and in depth interviews with 32 users, and social network analysis of 45,672 posts. The results reveal that LLMs act as new algorithmic gatekeepers, concentrating coverage on technology and entertainment, and acquiring a degree of intermediation (0.342) higher than that of traditional media. Users develop lay theories characterized by ambivalence between perceived efficiency and distrust of algorithmic selection criteria. The Bayesian regression model identifies engagement as the strongest predictor of change in the thematic agenda (β = 0.412). It is concluded that LLMs have structurally transformed gatekeeping and agenda-setting processes, requiring an update of classical communication theories to incorporate algorithmic mediation.