Technological advances in the field of artificial intelligence (AI) have reconfigured the modes of reception, circulation and authorial production, generating new tensions between originality, creativity and discursive appropriation. This study investigates the impacts of AI on authorial practices and meaning-building processes, considering how predictive and generative algorithms influence audience formation and discourse dissemination. From the Bakhtinian perspective, we examine the dialogicity present in the textual production mediated by AI and the epistemological implications of this mediation for the concepts of authorship and polyphony. Methodologically, a discursive analysis of texts generated by AI and academic debates on the subject is carried out, observing the relationship between algorithmic predictability and construction of meanings. The results point to a shift in the traditional conception of authorship, evidencing the need for new analytical categories to understand the contemporary discursive production.