Abstract
<jats:p>This article examines the transformation of media texts under the influence of generative artificial intelligence models. The relevance of this work stems from the rapid adoption of neural network technologies in news production, which is fundamentally changing the pragmatic and stylistic contours of media discourse. The aim of the study is to identify key linguistic and structural differences between news texts created by professional journalists and texts generated by the GPT-4o language model. A corpus of 200 news articles on socio-political topics served as the research material.The study utilizes a comprehensive medialinguistic approach, combining corpus analysis, stylistic tagging, and discourse analysis. The study found that AI-generated texts exhibit a high degree of syntactic standardization, reduced lexical variability, and a lack of authorial pragmatic markers. At the same time, specific patterns of metaphorical "hallucinations" were identified.</jats:p>