Creative agency on generative AI music platforms: Reframing artistic identity and the creative subject

Abstract

Generative AI music platforms promise to facilitate creative production without advanced technical skills. This study analyzed an approximately 22,000-word corpus of interface texts, terms of use, marketing materials, and blog posts from Suno, Udio, AIVA, and Soundraw using Fairclough’s Critical Discourse Analysis and Lakoff and Johnson’s conceptual metaphor theory. It examined how platform texts construct power relations and attribute agency. Three tendencies emerged. The agency game denotes the tension between the creative agency attributed to users in interface discourse and the contractual limits imposed through licensing and ownership clauses. The illusion of transparency denotes the backgrounding of algorithmic operation behind metaphors of magic, tool, and partnership. The rhetoric of democratization reflects the ideological function of the “anyone can be a musician” framing, which may bear on how professional expertise is valued. Disclosure of training material was assessed at three levels of specificity. Three analytical units on two platforms described in general terms where training material was obtained, one under litigation and two as a commercial guarantee; none identified specific works or datasets, and none disclosed any consent, attribution, or compensation mechanism for the relevant rights holders. The findings indicate that these platforms do not eliminate creative agency but render it discursively ambiguous by offering users a symbolic creative subject position while obscuring the distributed agency underlying production.