Ethics – June 2025
What do we talk about when we refer to the ethical value in our “Music 360 framework to quantify and qualify the value of music”? How could we convert it to real data on Music 360 digital platform?
The ethical value of music is an interesting topic that different authors have analysed. At UPV Team, we study it from two different dimensions:
- The ethics of the music ecosystem, along the value chain.
- The ethics in the transmission of some values, through their lyrics content.
How lyrics could transmit positive behaviours has been explored by Higgins (2023) and Preniqi et al. (2023), and it is a question on which the music ecosystem agrees. For example, we could find an E label on a song if it has “explicit” content such as strong language, mature themes, violence, sex or drugs. Anyway, labels and artists could enhance more social and ethical content.
In relation to how the music ecosystem can add ethics along its value chain, concepts such as transparency, copyright, compensation, support to local and independent labels or vulnerable groups, connect to the studies of Weijters et al. (2014) or Green et al. (2016). But, at that moment, the music ecosystem was less digital… what happens now with music platforms and generative artificial intelligence? They bring advantages and disadvantages to music artists.
On one hand, IA can provide support for detecting copyrighted content and balance the dependence of independent artists on streaming platforms. Music360 project coordinator, Jaap Gordijn, participated in the EKIP Policy Lab during Tallinn Music Week, where it was assumed that “Ethical AI in this context refers to algorithmic systems that promote fairness, transparency, diversity, and accountability in how music is recommended and surfaced to users… ensuring that AI-driven recommendation systems do not disproportionately favour major-label content or reinforce existing biases in language, genre, or geography… This could also include requiring platforms to offer user-customisable recommendation settings or to implement discoverability quotas for minority-language or independent music” (Siil et al, 2025).
On the other hand, music professionals do not feel so left out by GenAI because it is only used at certain stages of the creative process. They think that GenAI will impact general work or jingles on-demand for the audiovisual sector, but not more complex pieces where GenAI could support them. However, they feel the process needs to be more transparent, especially when using copyrighted materials. According to Batlle et al (2023), transparency “acts as an umbrella for multiple strategies and methodologies, incorporating explainability, interpretability, documentation, reproducibility, auditability and traceability”.
Some GenAI companies have acquired the rights from content creators or have signed up trust certificates, but the source of the material used is not always known. The legal cases of generic platforms such as Suno and Udio have created a debate on using copyrighted content to train the models without compensation for the authors, even if the result could be different from them. For Nayar (2025) two schools of thought arise: on the one hand, “technologists argue that LLM training is similar to human processes and therefore not considered infringement” and, on the other hand, these platforms are “only able to generate music because (they) sources (their) inputs from the work of other artists’ copyrighted music”.
This implies that copyrighted inputs need permission and compensation, while outputs would be transparent (CISAC and PMP Strategy, 2024). Some companies, such as the Spanish BMAT, work to preserve the authorship of copyrighted music in this digital environment. They have recently joined the DDEX (standards) board of members. We can notice how crucial standardised metadata are, because they could provide information to other actors that protect copyrighted music. However, these different initiatives may join in a single standard model.
Music360 