Shadow AI in Journalism: Between Ethics and Algorithmization

Abstract

This article examines the phenomenon of shadow AI in journalism, understood as the unauthorized or undisclosed use of generative artificial intelligence tools in editorial practice. Its central argument is that journalists’ covert use of tools based on large language models (LLMs) generates tensions and weakens their sense of professional agency over the content produced. The article draws on the author’s original research conducted among Polish journalists in the first quarter of 2025, as well as on selected international reports. It demonstrates that the absence of editorial policies and precise regulations, combined with ambivalent attitudes toward AI, fosters hidden practices of generative tool use. Consequently, shadow AI can no longer be viewed solely as a cybersecurity issue, with which such practices are most commonly associated. In light of the findings and the author’s analysis, the phenomenon extends beyond organizational concerns and emerges as a significant challenge to transparency, professional ethics, and the credibility of the media as institutions of public trust.