In the past summer, I have spent a bit of time experimenting with Suno, an AI music generation tool. This application turns short text prompts into complete songs. This especially interested me because I enjoy playing instruments, but have no experience in the creation of digital music. This new application makes it very easy to create melodies, harmonies, and lyrics that sound coherent. Suno and other comparable applications lower the barriers to music production, making music production accessible to individuals with no prior experience in creating music.
The ease with which these applications can now be used reflects a broader transformation in how music is composed and experienced. Briot (2020) explains that advances in deep learning have allowed AI models to learn musical structures from large collections of data and to generate new music that fits within those styles. He draws a distinction between two types of generation. On the one hand autonomous generation, where the system creates music on its own, and on the other hand composition assistance, where the user guides the process through creative input and feedback. In my opinion my experience with Suno fits this second description best. Because the tool does not replace creativity completely. It translates my prompts into musical output. In the way I used Suno, the AI application acted more as a creative assistant but sometimes also as an autonomous generator.
Apart from the fact that these new applications are very useful for people with no musical knowledge to suddenly create music, they also come with significant legal and ethical challenges. Gervais (2019) notes that copyright law is based on the concept of human authorship. Since AI operates mostly autonomously, it falls outside traditional legal definitions of creativity and ownership. This legal challenge became clear to me when I tried sharing an AI-generated track to Soundcloud and was asked to confirm the ownership rights. Although I had shaped the prompts and creative direction, I could not confidently claim to be the legal author of the music.
Hugenholtz and Quintais (2021) argue that creativity in copyright law involves three stages: the conception of an idea, its execution, and the final refinement of the work. Copyright protection, they note, requires meaningful human input across these stages. When an AI system carries out most of these steps autonomously, without relevant human creativity, the result cannot be regarded as a protected work. In my experience, a tool like Suno automates much of this process: the user provides a brief prompt, but the system composes, arranges, and polishes the final piece. As a result, the main creative labour lies with the algorithm rather than the human user. This not only limits the legal protection but also raises ethical concerns about authorship and artistic responsibility. It reinforces my earlier point that while applications like Suno make music creation remarkably easy, they also blur the boundaries of what it means to be creative human being.
Sources:
Briot, J. (2020). From artificial neural networks to deep learning for music generation: history, concepts and trends. Neural Computing and Applications, 33(1), 39–65. https://doi.org/10.1007/s00521-020-05399-0
Gervais, D. J. (2019). The machine as author. Iowa Law Review, 105(5), 2053–2106. https://ilr.law.uiowa.edu/print/volume-105-issue-5/the-machine-as-author/
Hugenholtz, P., & Quintais, J. P. (2021). Auteursrecht en artificiële creatie. Auteursrecht, 47–52. https://pure.uva.nl/ws/files/61822465/Auteursrecht_2021_2.pdf
