In 1982, a touring Barry Manilow began utilizing synthesizers to replicate string and horn parts during performances in the United Kingdom. The UK Musicians’ Union responded with a ban on synthesizers and drum machines, citing concerns about replacing human musicians. A U.S. union representative reportedly commented on the situation at the time.
Current legal considerations regarding AI-generated music present similar questions. Generative AI systems are now capable of composing songs, and these compositions are achieving commercial success.
U.S. copyright law differentiates between musical compositions, which are the organization of musical ideas, and sound recordings, which are specific sequences of sounds. Copyright protects the expression of ideas, not the ideas themselves. Public domain material, licensed content, and common musical elements are not protected. A work must be fixed in a tangible medium, such as through notation or recording, and must demonstrate originality.
The standard for originality is minimal, requiring only that the work not be copied from another source. Traditionally, musical originality has been assessed based on elements like melody, rhythm, lyrics, and harmony, although evaluation is case-specific. Originality is not automatically established, even in sound recordings, and requires a substantive contribution from the author.
Past legal interpretations emphasized melody in determining musical originality. Modern perspectives are more inclusive; for example, the Monkees’ song “Zilch,” consisting solely of spoken words in polyrhythmic patterns, has been recognized as a copyrightable musical composition.
The question of authorship extends to non-human creators. In 2018, the United States Ninth Circuit Court of Appeals determined that photographs taken by a crested macaque named Naruto were not eligible for copyright protection, as authorship is legally reserved for human beings.
This principle was reaffirmed in a 2025 decision by the United States Court of Appeals for the D.C. Circuit, Thaler v. Perlmutter, concerning AI-generated works. The court based its decision on the language of the Copyright Act and did not address whether the U.S. Constitution’s clause regarding author’s rights might apply to AI-generated music.
The AI program itself and its outputs are distinct entities. The outputs are considered machine responses to prompts, not works authored by the program’s creator. Basic prompts, lacking sufficient originality, cannot be copyrighted, and do not establish a work-for-hire relationship. For example, a prompt like “write a funny medium tempo song for voice and piano in a major key about my dog” would not confer ownership of the resulting song.
The Copyright Act also prohibits copyrighting short phrases, titles, or slogans. While sufficiently original instructions may be copyrightable, the copyright applies to the instructions themselves, not to the resulting musical work.
John Cage’s composition “4’33””, consisting of silence, provides a case study in the boundaries of musical expression and copyright. The work’s score consists of a tempo indication and instructions, leaving the performance and ambient sounds to chance. Cage’s copyright likely extends to the text of his instructions, but not to performances of silence. Any copyright in a recording of the piece would require creative input from a producer or engineer in the recording process.
Spotify and Universal Music Group Announce AI-Powered Music Creation Tool
Pete Townshend and Primary Wave Music Establish Partnership
What InterSpace’s New Metadata Policy Means for Your Releases
