New journal article: RAIM: Operationalizing Trustworthy AI in Music Generation
New journal article: RAIM: Operationalizing Trustworthy AI in Music Generation
de Berardinis, J. et al.
Generative AI is radically reshaping the musical landscape, unlocking new forms of creative expression while simultaneously raising profound ethical, legal, and societal concerns. Key among these are the potential displacement of human creativity, copyright infringement stemming from vast training datasets, and the lack of transparency, explainability, and fairness mechanisms. As generative systems become pervasive, shifting from reactive regulation to responsible design is crucial. While previous work has tackled isolated aspects of generative systems, we propose a systematic framework grounded in the Ethics Guidelines for Trustworthy AI produced by the European Commission’s High-Level Expert Group. Focusing on generative music AI, we contextualize these requirements into 45 responsible features that address trustworthiness across technical, legal, and artistic dimensions. We evaluate this framework through a multi-stage Delphi study involving a global panel of experts from academia, industry, and regulatory bodies, comprising AI ethicists, legal experts, music AI practitioners, and artists. The results demonstrate a high degree of expert endorsement for the proposed features, while revealing thematic tensions between the protectionist priorities of artists and the technical constraints of practitioners. We argue that the framework functions as a translation layer, operationalizing ‘complementarity’ to harmonize different stakeholder values rather than forcing a monolithic consensus. Ultimately, this work provides an empirically grounded foundation for the design, evaluation, and regulation of Responsible AI in Music (RAIM). The manuscript is accompanied by a website: https://casbahlab.github.io/raim-framework/ link