The research aims to explore the ability of GPT-4 to emulate the style of two Italian cultural magazines active in the 1960s, «Quaderni Piacentini» and «Quindici». Using a corpus derived from the early issues of these magazines, the study assesses whether GPT-4 can bypass stylometric analysis by producing text that reflects the editorial strategy of a specific magazine. After using GPT-4 for generating emulative texts, a stylometric analysis was conducted to compare the AI generated texts with the original corpus. Comparison with traditional stylometric methodologies has allowed the identification of aspects where the two journals diverge, and consequently, the nodes on which the model focuses for stylistic and thematic differentiation. The research intends to open new applications on the use of stylometry for the computational analysis of texts related to specific cultural contexts; indeed, until now, the scientific community has focused on the ability of LLMs to faithfully reproduce the styles of different authors. Applying these methodologies to the field of magazines would allow for broader considerations on editorial strategies, reading trends, and the processes of idea circulation.