Song Created with Artificial Intelligence: Who is the Author?

The rise of Artificial Intelligence (“AI”) in music has rocked the copyright landscape. Before the introduction of AI, song creation was inseparable from human creativity; there was the composer crafting melodies, the lyricist shaping words, the arranger refining the sound, the singer delivering emotion, and the producer bringing the final piece to life. Today, however, a song can be created with a simple prompt such as “create a sad Indo pop song with the voice of a baritone and lyrics about loss”. In a matter of seconds, the AI system will be able to create a melody, lyrics, and arrangement, even providing vocal performances that closely resemble human singers.

This phenomenon raises a major question: Who is the author? Is it the person who typed the prompt? The AI developer? The AI itself? Or could it be that no copyright arises at all? 

Under the current Indonesian Copyright Law, for a work to be copyrightable, it must be unique and personal, originating from the inspiration, ability, thoughts, imagination, skill, or expertise of its author. The law further defines an author as a person or a legal entity, which excludes AI as it is considered neither. Therefore, if a song is created entirely by AI without sufficient human involvement, it may not be eligible for copyright protection and the right holder status remains ambiguous.

However, in the scenario where human participation plays a substantial role in the creative process, for example, by writing or refining the lyrics, determining the song’s structure, editing the AI-generated output, or carrying out the mixing, mastering, and overall artistic development of the song – in such cases, it is likely that the human contribution may be sufficient to establish authorship, at the very least, ownership rights over the final product.

The ongoing discussions surrounding Indonesia’s proposed Copyright Bill suggest that regulators are beginning to adapt the legal framework to the realities of AI-assisted creativity by expanding the definitions of “Work” and “Author” to include creations produced with the assistance of AI. However, protection would still depend on the existence of meaningful human contribution. Elements such as conception, curation, refinement, artistic direction, and the ability to demonstrate a documented creative process would remain central in determining whether a work qualifies for copyright protection. Consequently, an entirely autonomous AI-generated output without human contribution may not qualify for copyright protection. 

This position was also reflected in statements made by Professor Ahmad Ramli during the public hearing of the Copyright Bill. He emphasized that AI-generated content must be positioned as a tool rather than as an Author in its own right, reaffirming that copyright protection continues to depend on human creative contributions.

Considering the above, under both the current Copyright Law and the proposed Copyright Bill, the “Author” of an AI-assisted song is likely to remain the human who provides meaningful creative contribution, while AI itself is positioned merely as a tool in the creative process.

AI is rapidly transforming the music industry, creating a conundrum for industry players, even regulators. On one hand, AI has democratized music creation, given rise to what many now call “instant musicians” – individuals who has little to no ability to play instruments, yet are capable of creating a song. It can also serve as a valuable tool for professional musicians, particularly in developing demos and experimenting with new ideas. On the other hand, there is growing concern that AI is trained using millions of copyrighted songs without the right holder’s permission.

Ultimately, the challenge presented by AI is not simply choosing between innovation and copyright protection, but about finding balance between the two. While creators deserve protection for the works used to train AI systems, excessive restrictions may also hinder innovation and limit public access to creative technologies. After all, the public is not merely a consumer of music, but also a key driver of the commercial value of copyrighted works. As AI continues to evolve, the legal framework must therefore balance the interests of rights holders, technological development, and the broader public. 

Yet beyond this delicate balance, a deeper question emerges: does a song still have a soul when its creator is an algorithm?