Research on the Ethics and Governance of Generative Artificial Intelligence Music Creation

[PRE-RELEASED VERSION]

Linlin Peng

Linlin Peng is postdoctoral researcher and Associate Professor in the Piano Department of the Music College at Hunan Normal University; supervisor of Master’s students in the programmes “Piano Performance” and “Music Pedagogy.” Serves as a national expert on the assessment of arts specialisms under the Ministry of Education of the People’s Republic of China.

ABSTRACT
The aim of this study is to analyze the ethical dimensions of using generative artificial intelligence (AI) in music creation, taking into account the risks of cultural appropriation in the context of Chinese heritage, and to develop regulatory mechanisms that balance innovation with human creativity. In China, there is a state agency called the Cyberspace Administration of China (CAC), which regulates cyberspace and the digital environment, monitors online information flows, and develops and enforces regulatory acts governing the field. Despite the growing number of technical developments and regulatory proposals in the area under study, there remains a substantial gap in empirical research examining how the professional community itself perceives these issues. This article aims to fill that gap by addressing the following research question: How do experts in music, technology, and law in China assess the ethical risks of using generative AI—particularly cultural appropriation—and which regulatory mechanisms do they consider most effective for minimizing those risks? The study was conducted within the qualitative research paradigm, with elements of descriptive quantitative analysis and anchored in an interpretivist approach: expert surveys were conducted with 23 specialists from China (musicologists, AI developers, lawyers, cultural studies researchers) using a custom-designed questionnaire consisting of two blocks of questions. The analysis included thematic grouping of responses and calculation of respondents’ agreement percentages. Key findings: substantial progress in transformer-based models and generative adversarial networks (GANs), but significant risks of appropriation (95% of respondents) and loss of authenticity; practical recommendations were developed based on the identified expert consensus.

KEYWORDS
Generative AI; ethics in music; cultural appropriation; CAC regulations; intellectual property; cultural safety