Drawing on Systemic Functional Linguistics (SFL), this study explores the (con)textual capabilities of ChatGPT (4o), one of the most popular contemporary artificial intelligence chatbots based on large language models, in relation to the contextual system of SOCIAL DISTANCE, which describes the degree of "intimacy" realized by the text. To do so, we worked with two parallel corpora: in one, ChatGPT was instructed about the system and its semiotic realization; then, we asked it to paraphrase five natural texts according to the four degrees of social distance described by the system and following the proposed realization principles. In the other, we simply asked it to paraphrase the same texts according to degrees of "intimacy," "personality," and "formality" – emic terms that often refer to different degrees of SOCIAL DISTANCE. The artificial paraphrases were then qualitatively analyzed using SFL lexicogrammatical and discourse semantic descriptions. The results suggest that the explicit instruction on the principles governing the system's realization led to paraphrases with more frequent and salient realization indices for each degree of SOCIAL DISTANCE, with relevant implications both for the analytical investigation of artificial productions as sociosemiotic objects and for applications in different areas.