The Weakened Frontiers: Understanding Legal Challenges In Generating Music Using AI

Written By-Chandril Chattopadhyay & Satyaki Paul

“I know it's true, it's all because of you,And if I make it through, it's all because of you”-John Lennon

Nothing can be better than starting this article with the lines from the song “Now and Then” composed by John Lennon.

The Beatles made an everlasting imprint on music history. Their timeless melodies, inventive compositions, and exquisite lyrics have captivated audiences all over the world. However, John Lennon's tragic death in 1980 left fans lamenting the loss of one of the band's most prominent voices.

Though not confirmed “Now and Then''  as a final Beatles song is going to be released this year; The Beatles member Paul Mccartney has confirmed. In an interview with the BBC, he mentions that AI Technology has been used in a manner to “ extract” Lenon’s voice from an old demo. This could never be possible without the existence and years of tremendous progress in machine learning and Artificial Intelligence (AI). Advances in artificial intelligence have opened up a wide range of new possibilities, even allowing us to investigate the possibility of extracting John Lennon's voice to produce a last Beatles song that embodies his essence. Have we ever expected that there still can be a song produced by The Beatles?

If we shift our pov to India, a great example of music converging with AI would be the song called “Clash”. Loved by millions of people this alternative version from the original version by Diljit Dosanjh actually carries the voice of Sidhu Moose Wala, one of South Asia's most influential Hip Hop personalities who was shot dead by gunmen on 29th May 2022. Artificial Intelligence is being used to recreate Sidhu Moosewala’s voice which is being used to recreate several Punjabi tracks. Here AI can be seen to be used to retain someone’s voice in order for it to sustain through the generations. But what about the ethical issues? What about the legal frameworks? What about the fact that the music industry is being economically exploited due to the advent of AI-generated music?   

Unleashing Creativity: The Process of AI Music Making:

In recent years, the music industry has witnessed a significant shift with the emergence of Artificial Intelligence (AI) technology. AI-generated music, a remarkable application of machine learning and deep neural networks, is gradually taking over the music scene.

AI-generated music has ushered in a new era of limitless creativity. AI computers can learn patterns, structures, and styles across genres and time periods by analysing massive volumes of existing musical material. These algorithms may then develop fresh songs that mirror the features of specific musical genres, or they might blend various genres to create something entirely new.

Composers, producers, and even music enthusiasts are embracing the possibilities offered by AI, allowing for the creation of innovative and unique musical pieces. Thus it is imperative to elucidate and throw light on how AI-generated music is revolutionising the contemporary music scene across the globe, both creatively and technologically.

The introduction of AI-generated music has also transformed collaboration within the music business. Musicians may now use artificial intelligence systems to explore new ideas and views. AI algorithms can produce musical fragments or themes that musicians might use to expand into whole works. This collaborative technique has the ability to ignite creativity and drive musicians to explore previously unknown musical worlds, resulting in the development of distinctive and appealing music.

Not to mention the serious and vigorous manual labour that was being put to create music has been reduced greatly by the advent of AI. Composing music traditionally entails hours, days, or even weeks of human labour. AI can speed up the time-consuming components of music composition, such as melody development, chord progressions, and even arranging and mixing. There is no doubt regarding the excellence of AI in the music scene. A deluge of AI music content is being made from every part of the world. Imitating the voice and even the mannerisms of the singers and the artists and voila what you get is a piece of music which closely replicates the artist of your choice within moments of wait.

Unveiling the Challenges:

The advent of AI-generated music has brought forth a set of challenges and concerns for the music industry. While AI-generated music presents exciting opportunities for artistic expression and creativity, it also poses challenges in the realm of music licensing.

"A work that contains no human authorship is not eligible in any categories."

This was the statement by the Recording Academy of the world’s most recognised music award Grammy. To be nominated, musicians now must contribute at least 20% of their work to an album. Previously, any producer, composer, engineer, or featured musician on an album may be nominated for album of the year, even if they had a small role.

AI has no doubt sped the creation of music. Apple Music, Amazon Music, and Spotify now have over 100 million songs. It would take hundreds of years to listen to them all. More have been posted on SoundCloud. AI technologies are democratising music creation. However, there is a risk that a deluge of AI-generated content may be released onto streaming platforms, vying for your attention with actual people and their compositions.

Lack of Originality and Copyright Ownership:

One significant problem faced by the industry is the question of originality and authenticity. As AI algorithms learn from existing musical compositions, there is a risk of generating music that closely resembles existing works, potentially infringing upon copyright laws. This raises issues of intellectual property and artistic ownership, as well as the potential devaluation of human creativity in the music-making process.

Royalty Distribution:

Furthermore, the rise of AI-generated music also poses challenges regarding fair compensation and royalty distribution, as the traditional model for determining music rights may not be adequately equipped to handle the complexities of AI-generated compositions. The music industry is largely pushing against AI. Universal Music Group, one of the biggest names in music production, has become the strongest voice against the advent of AI-generated music in music listening platforms.   

Attribution and Recognition:

A work that is otherwise infringing may be considered fair use if it serves some type of commentary or criticism, such as the parody in the classic case Campbell v. Acuff-Rose Music. While parodies have a commercial motive and provide a comparable function to the original, their transformative nature and critique classify them as fair use. AI music also substantially transforms and works on the existing music. But it does so without the intent of commenting or criticizing or making any reference to the original work. AI music works on the already produced works and takes a free ride to create traction and commercialise over it. The work most of the time directly competes with the original work, in turn, it reduces incentives.

Sampling and Plagiarism:

Artificial intelligence algorithms imitate numerous musical styles and genres, prompting worries about sampling and copying. Because AI systems learn from existing music, AI-generated songs may mistakenly imitate copyrighted components. When licensing AI-generated music, leads to serious legal difficulties because the original authors may claim infringement.

India is a diverse country with rich cultural and regional musical traditions. AI algorithms trained on a predominantly Western-centric music dataset may not accurately capture the nuances and cultural variations of Indian music. Licensing AI-generated music that incorporates Indian musical elements requires careful evaluation and negotiation to respect the cultural heritage and regional rights associated with traditional Indian compositions.

Digitised music generation: A walk down the timeline

In the Sixty Eighth Annual Report of the Register of Copyright, U.S. for the Fiscal Year 1965, there was a significant question posed before the Office pertaining to whether a “ work” of the computer in generating music can amount to authorship for the human or of the computer,  The Copyright office has in fact received an application for registration of such musical work.Therefore digitally created music has been a significant phenomenon in the copyright domain even a few decades back.

Determination of whether an AI output is original or copyright-protected is often difficult to gauge. There is a high possibility that the economic actors can deviate from a legally accurate attribution of rights which is fact specific. Thus with control, there is often negation of the responsibility of those in the field of copyright and hence the claiming the copyright becomes easy. This is also possible when the copyright laws say otherwise in terms of originality. The false authorship in terms of the AI output shall make the discourse much more complicated than it already is.

When Plato was writing Republic, he insisted on making it a point that art around us is a mimesis or imitation of what exists as a form. But, when AI generates existing music on some already existing and copyrighted music, do we call it a mimesis? How is an AI generated music twice removed from the existing music? The answer is still in the grey area and needs further deliberations to make it understandable.However, in compliance with what Emile Zola talks about imitated art, the experimenter goes beyond the spectrum of existing music to create deviations and make it known to the world. So do the new music

The Governments have acted differently when it comes to holding the bridle of the horses that have rampantly gone on to create music with AI bots without licensing the music from the copyright owners. The UK Government has shown a  pro stance when it comes to helping the tech companies in making such “ generated music” on top of existing music and this has been mostly done at the expense of the Music labels. However, such legislations shall be violative of the Berne Convention and be against the international law that exists with respect to protecting the IP rights of the music.

The author of this piece used Mubert with the prompt: “make an electronic music based on a Bengali folk song” and the existing results come as unique and shocking as it bases the new track on some existing folk song whose copyright is perhaps owned by some folk artist. The tracks can be downloaded and voices of artists can be used to make a completely new track of its own. When AI shall be training data, the second kind of infringement shall happen.

Thus, AI-generated music frequently employs a continuous output paradigm, resulting in an infinite stream of songs. This calls into question the scope and length of music licensing contracts. It is critical to define the scope of usage covered by licences and if a single licence may cover the full output or if multiple licences are necessary for each piece of AI-generated music. It is critical to create adaptive licensing models that take into account the particular qualities of AI music. Collective Management Organizations or CMOs are critical in managing music licensing and collecting royalties for artists. The growth of AI-generated music, on the other hand, poses issues for these organisations. Identifying, recognizing, and appropriately crediting royalties to AI music producers can be difficult. To build successful procedures for controlling and distributing royalties in AI music, CMOs, AI developers, and industry stakeholders must work together.