By
Stuart Kerr, Technology Correspondent, LiveAIWire
In April 2023, a track called “Heart on My Sleeve” —
featuring what sounded like Drake and The Weeknd — went viral on TikTok
before being removed at Universal Music Group’s request. The track had been
created by an anonymous producer using AI voice cloning. No sample had been
taken; no existing recording had been used. The AI had learned the vocal
characteristics of two of the world’s most commercially successful artists
and synthesised new performances from scratch. Universal’s lawyers called it
a copyright violation. Technologists called it a demonstration of what was
now technically possible. Both were right.
The music industry is confronting an AI moment more disruptive
than streaming, more legally contested than sampling, and more
philosophically challenging than any previous technological shift in its
history. The questions it raises — about authorship, compensation, cultural
authenticity, and the economics of creative labour — are not resolvable by
copyright law alone.
What Generative AI Can Do With Music
The range of AI music capabilities now available, through consumer
tools and research systems, is broad. AI can generate original compositions
in specified genres and moods from text descriptions. It can clone vocal
performances with sufficient accuracy to produce new material
indistinguishable from the original artist. It can separate audio tracks into
constituent elements and recombine them in new configurations. It can
harmonise, arrange, and produce at the level of a competent human session
musician, and in some stylistic domains it exceeds that
level.
Tools like Suno, Udio, and Google’s MusicLM have made music
generation accessible to anyone with an internet connection. The barrier to
creating a professionally plausible musical composition has fallen from years
of training and thousands of pounds of studio time to a text prompt and a few
seconds of processing. This democratisation of music creation is genuinely
significant for people who have musical ideas but not the technical skills to
realise them. It is also genuinely significant for the session musicians,
composers, and producers whose livelihoods depend on being paid for those
skills.
The Copyright Battlefield
The legal questions raised by AI music generation are being
contested simultaneously in courtrooms, legislatures, and industry
negotiations. The most fundamental is whether training AI models on
copyrighted music without licence or compensation constitutes infringement.
Lawsuits brought by Universal Music Group, Sony Music, and Warner Music Group
against AI music companies Suno and Udio, filed in 2024, will test this
question in US federal court. The outcome will establish precedent for AI
training on copyrighted material across creative industries.
The secondary question is whether AI-generated music itself can be
copyrighted, and by whom. The US Copyright Office has concluded that works
created entirely by AI without human creative input are not eligible for
copyright protection. Works in which AI is a tool in a human creative process
may qualify, depending on the degree of human authorship involved. The
precise boundary is contested and will be litigated in cases whose facts are
still accumulating.
A third question concerns the use of an artist’s voice or
stylistic signature without consent. Current US copyright law protects
specific recordings but not an artist’s vocal characteristics or musical
style in the abstract. Several US states have passed or are considering right
of publicity legislation that would extend protection to AI voice cloning;
federal legislation has been proposed but not enacted. The NO
FAKES Act would create a federal right against unauthorised digital
replicas of individuals including voice and likeness.
The Economics of AI Music and Human Musicians
The economic impact of AI music generation on working musicians is
not yet fully measurable, but the directional pressure is clear. Session
musicians who record background tracks, composers who create production music
for advertising and media, and producers who specialise in genre-specific
commercial work are all in roles where AI substitution is already technically
feasible and economically attractive for commissioning
clients.
The musicians most insulated from AI competition are those whose
commercial value lies in their personal identity and cultural presence rather
than their technical musical output. A Taylor Swift concert is not
substitutable by an AI-generated performance, because what fans are paying
for is not simply music but the specific cultural experience of Taylor Swift.
The demand for that experience may even increase as AI-generated music floods
the market and human authenticity becomes a scarcer and therefore more
valuable signal.
The musicians most exposed are those in the middle and lower tiers
of professional music: the session players, the library music composers, the
working producers who create the musical infrastructure that surrounds the
stars. Research from the Musicians
Union in the UK has documented members reporting lost commissions
to AI-generated alternatives, particularly in advertising, gaming, and
podcast production contexts where the brief prioritises functional music
rather than artistically distinctive work.
AI-Assisted Creation and the Question of
Authenticity
Not all AI music is AI-replacing-human music. A significant
proportion of the most interesting AI music work involves human artists using
AI as a compositional tool — generating material to react to, exploring
sonic spaces they could not have accessed manually, and using AI processing
to achieve effects that expand rather than substitute for their creative
agency. Artists including Holly Herndon, Grimes, and Arca have engaged with
AI as a creative collaborator in ways that complicate the replacement
narrative.
The authenticity question is ultimately cultural rather than
technical. Music cultures have always negotiated what counts as authentic
creative expression: the introduction of electric instruments, drum machines,
sampling, and Auto-Tune all generated debates about authenticity that were
eventually resolved through cultural acceptance of the new tools as
legitimate means of expression. AI will likely follow the same trajectory,
with the period of contestation resolved by the emergence of new genres and
practices that are understood as distinctively AI-assisted rather than either
purely human or purely automated.
The connection to the
broader cultural response to AI in creative fields is direct: music
is the domain where the economic and cultural stakes of AI creativity are most
immediately visible, because the market for music is large, the affected
workforce is numerous, and the cultural significance of music in human life
is profound enough to make the question of who creates it genuinely matter to
people who are not themselves musicians.
The music industry is not going to resolve the AI challenge
through copyright litigation alone. It will require new licensing frameworks,
new compensation models, and new cultural norms about the relationship
between AI generation and human creativity. The industry that invented
sampling — and eventually developed a workable licensing infrastructure for
it — has the institutional experience to develop those frameworks. Whether
it will do so before AI-generated music reshapes the market to a degree that
makes negotiated settlement less attractive is the question that the next few
years will answer. For context on how AI
is raising similar consent questions in the context of deceased
artists, the parallel with voice cloning of the living is
uncomfortably close.
The music industry’s negotiation with AI is also a negotiation
about the future of creative labour more broadly. The precedents set in music
licensing, collective bargaining agreements, and copyright litigation will
influence how AI’s relationship with creative workers is governed across
film, television, publishing, and visual art. The creative industries are, in
this moment, the testing ground for questions about AI and human labour that
will eventually apply across a much wider range of knowledge work. How well
creative workers organise, how effectively they mobilise legal and political
support, and how successfully they articulate the value of human creative
contribution in an AI-generation context will determine not only their own
futures but the template for worker responses to AI across the economy. The
broader question of who bears the costs of AI-driven economic
disruption is nowhere more visibly contested than in the music
industry right now.
About the
Author
Stuart Kerr is a technology correspondent at
LiveAIWire, covering artificial intelligence, emerging technologies, and
their impact on society and industry.