In April of 2023, a mysterious track titled “heart on my sleeve” appeared online: a duet seemingly between Drake and The Weeknd, except neither star had anything to do with it.

The song was generated by an anonymous producer using artificial intelligence to clone the artists’ voices, and it racked up over 15 million views on TikTok within days.

By the time Universal Music Group (UMG) and streaming platforms scrambled to pull it down, the so-called “Fake Drake” track had already proven a sobering point: AI had arrived in pop music, and the traditional gatekeepers were caught unprepared.

This was no longer a sci-fi hypothetical or a gimmick confined to tech demos. It was a viral hit made by software, and it sounded convincingly real. The alarm bells rang across the global music industry, from boardrooms in Los Angeles and London to lawmakers in Washington, signaling that a profound disruption was underway.

As it turned out, this was simply the opening note of a seismic shift still unfolding.

The Rise of AI Music Platforms

In the two years since, AI music has undergone a dramatic transformation. What began as a viral stunt has evolved into a venture-backed ecosystem, with platforms like Suno and Udio offering tools that let anyone generate full songs with vocals, lyrics, and production in seconds. The technology that once shocked the industry now threatens to reshape it from the inside out.

Their ascent has been swift and staggering. Suno, founded in Massachusetts, raised $125 million in venture funding in 2024, climbing to a reported $500 million valuation. Its competitor Udio secured backing from Andreessen Horowitz, a clear signal that Silicon Valley sees generative AI as the next frontier in music creation.

What makes these tools so powerful, and to many in the industry so unsettling, is how they were built. Their models were trained on vast, unlabeled libraries of existing music, with entire catalogs from legacy acts like The Beatles to global stars like Beyoncé effectively repurposed as raw material. With just a few words, Suno and Udio let users generate songs in the style of recognizable artists, often with vocals and production that verge on the indistinguishable.

v4.5 just dropped for Pro & Premier subscribers 🔥 A wider range of genres, richer vocals, & enhanced prompt understanding for songs that match your vision.

What’s New:
🙌 Expanded genres & smarter mashups: More genre options — Blends like midwest emo + neosoul or EDM + folk… pic.twitter.com/MGp3zH8VHd
— Suno (@SunoMusic) May 1, 2025

It’s not hard to see both the appeal and the danger. For aspiring musicians and content creators, AI lowers the barrier to entry dramatically. You no longer need a studio, instruments, or even musical skills to create a song. But for established artists and rights-holders, the technology can feel like looking into a funhouse mirror: their styles and even their voices cloned without consent, proliferating online in potentially infinite permutations.

The scale of AI-generated music proliferation has only intensified. Boomy, one of the earliest consumer-facing AI music platforms, confirmed that users had created over 14 million songs by 2023, accounting for around 13.8 percent of the world’s recorded music catalog.

Meanwhile, overall uploads to streaming platforms have surged. Industry heavyweights like Universal’s Sir Lucian Grainge and Warner’s Steve Cooper estimate around 100,000 new tracks are now added daily to platforms such as Spotify, Apple Music, and SoundCloud. AI-generated content appears to be a major contributor to that uptick: Deezer reports that 18 % of daily uploads (roughly 20,000 tracks) are fully AI-created, many of which are suspected bots used for stream manipulation. Spotify itself purged tens of thousands of Boomy-generated tracks in mid-2023 over concerns of “artificial streaming” and fraudulent activity. This is no longer a niche phenomenon but a pressing challenge for platforms.

“Any way you look at it, it’s over-supply… it’s simply bad for artists, bad for fans, and bad for the platforms themselves.”
LUCIAN GRAINGE

“Any way you look at it, it’s over-supply,” declared Grainge. “Whether or not AI creates [that over-supply], it’s simply bad – bad for artists, bad for fans, and bad for the platforms themselves.”

Grainge noted that earlier generations of AI music were low-quality and had “virtually no consumer appeal,” but the latest wave is far more sophisticated and much of it has been trained on copyrighted material. In other words, this is no longer a curiosity. It is a direct challenge to the music industry’s foundations.

Inside the Legal Showdown

As AI music startups gained traction, the music industry’s initial (and unsurprising) response was to reach for its legal arsenal. In mid-2024, the three major record companies – UMG, Sony Music, and Warner Music – filed lawsuits against Suno and Udio in the U.S., accusing them of wholesale copyright infringement.

The core issue: these AI firms had admittedly trained their models on “the majors’ recordings without permission,” ingesting countless copyrighted songs to learn how to compose and mimic.

“AI companies, like all other enterprises, must abide by the laws that protect human creativity… There is nothing that exempts AI technology from copyright law,” the labels argued in their complaint. Suno and Udio’s defense was that their training process constituted “fair use” under U.S. law – a claim the labels vehemently rejected.

The companies defended their actions by claiming “fair use,” likening training to a student learning from listening, rather than illegal copying. But the majors counter that AI, unlike a human, can “directly compete with, cheapen, and ultimately drown out” original artists—and that its outputs risk being near-identical replicas. This legal crossfire has thrust crucial questions into the spotlight: What counts as transformative learning? When does influence become infringement? And who owns the AI-generated song that sounds like a famous artist?

Independent artists have joined the fray. In June 2025, country singer Tony Justice filed a class-action suit claiming Suno and Udio had “used [his] recordings and works from thousands of class members without authorization,” accusing the platforms of outright theft rather than innovation. His lawsuit also cited a U.S. Copyright Office report stating fair use does not justify ungated training on creative works, especially when AI-generated content may undermine original market value.

The legal fight isn’t limited to American courts. In early 2025, GEMA—the German music licensing society—filed suit against Suno in Germany, reflecting parallel concerns under European copyright law. Lawsuits are also multiplying in other creative domains, from literary AI to image- and video-based deepfakes. Together, these cases could reshape copyright doctrine for the 21st century.

In a notable shift, the record labels have begun courting AI music platforms — even amid lawsuits. By mid-2025, all three majors entered confidential negotiations with Suno and Udio over licensing training rights, royalties, attribution standards, and even equity stakes. As one Bloomberg report noted, this pivot reflects a strategic blend of resistance and partnership: “If you can’t beat them, sign them”.

These negotiations aim to balance control and innovation. Labels seek stronger safeguards: artist opt-outs, pre-approved song use, headphone-level attribution, and fingerprinting systems akin to YouTube’s Content ID. AI firms, in turn, want flexible terms that keep innovation alive without breaking the bank. Their ambitions converge in detecting AI-generated tracks with 99.9% accuracy and sharing royalties when output mimics identifiable artists.

Even Grainge, once staunchly opposed to unbridled AI, has begun to articulate a kind of cautious optimism about “responsible” use of the technology: “We have to be ready. Let’s not be afraid. Let’s just be prepared.”

The coming months will determine whether this blend of legal pressure and private dealmaking can forge a sustainable middle path. Will training require transparent licensing? Will each AI-generated track require fingerprinted attribution? And can artists get royalties from sounds they helped inspire? As cases move through international courts and private negotiations progress, the answers will shape the future of music copyright in an AI world.

Artists vs. Algorithms

Understandably, many artists have approached AI music with a mix of anxiety and anger, seeing it as a threat to their very identity. Legendary singer-songwriter Sting captured that fear pointedly while accepting a songwriting award last year.

“The building blocks of music belong to us, to human beings,” Sting told the BBC, warning that “that’s going to be a battle we all have to fight in the next couple of years: defending our human capital against AI.” The 71-year-old rock icon dismissed algorithmic songs as emotionally hollow, comparing them to lifeless CGI films, and said that while AI might have a place in electronic music, it cannot replicate the emotional depth of a real song.

Sting’s stark “battle” rhetoric resonated with many fellow musicians who feel that something deeply personal is under siege. The idea that a faceless program could appropriate one’s voice or style – even after death, as AI can resurrect voices of late artists – is a chilling prospect.

“This misuse hurts artists and their fans alike,” testified Recording Academy CEO Harvey Mason Jr. to Congress, referring to AI clones of singers ranging from Frank Sinatra to Pop Smoke.

Drake, Bad Bunny, and other chart-topping artists have voiced similar disgust at unauthorized AI doppelgängers. For many performers, their voice and persona are extensions of themselves – assets that they have carefully honed and marketed. Having those cloned and manipulated without permission is, as Grainge put it, essentially “identity theft”.

Yet, amid the backlash, a counter-narrative is emerging from a subset of artists who see AI as an opportunity rather than an enemy. The most famous example is Grimes, the avant-pop musician who essentially open-sourced her voice in 2023 . Rather than fight the tide of deepfakes, Grimes launched an AI tool (called Elf.Tech) that allows anyone to create music using her voice model. She actively encouraged fans to experiment, announcing that she would split 50% of any royalties from successful AI-generated tracks.

“I’ll treat it the same as if you produced a song for me.”
GRIMES

“I’ll treat it the same as if you produced a song for me,” she said, framing it as a new form of collaboration. Grimes asked creators to be “tasteful” and reserved the right to block extreme uses, like a faux “AI Grimes Nazi anthem,” she joked. She views co-creation as part of her ethos, speaking of “killing copyright” and advocating for “communal voice ownership” and the blurring of artist and audience.

It’s a radical stance, but one that’s earned her praise for forward-thinking. And practically speaking, Grimes’ approach could turn AI from a threat into a new revenue source. If a fan-made Grimes-AI track goes viral, both she and the fan stand to profit.

Other artists and producers are cautiously dipping their toes into AI’s waters. The superstar producer Timbaland, after initially sparking controversy by using AI to mimic the late Notorious B.I.G. on a track, has since launched an AI startup of his own. His company is working on software to create custom AI voices, and Timbaland envisions it as a tool for artists to resurrect the voices of departed legends or simply play around with new sounds.

“We know this is the future,” the vocal Suno supporter said in interviews, emphasizing he wants to “embrace AI” but do it ethically, with respect for the original artists’ estates. Just last week, Timbaland was forced to apologize to producer K Fresh after using his song without permission in an AI demo for Suno, illustrating the challenges of navigating the fast-moving sector.

Even some veteran artists have found silver linings. Paul McCartney made headlines by using AI audio restoration to isolate John Lennon’s vocals from an old demo, enabling a final Beatles song, entitled “Now and Then,” to be completed in 2023. It won the Grammy for Best Rock Performance this year, offering a poignant example of AI as a creative aide rather than a counterfeit threat.

And across genres, many musicians are quietly using AI tools to assist in music production (for instance, AI-driven mastering, or generating ideas for melodies and beats). According to one survey, about 40% of artists have a favorable view of AI’s role in music and believe it can align with their creative goals. Particularly in electronic and hip-hop music, more than half of creators have already incorporated some form of AI into their process.

But even among the curious and open-minded, there’s an undercurrent of concern. In the Goldmedia survey of thousands of European composers, 71% said they fear AI could make it impossible for music creators to make a living. They worry about being “replaced” or drowned out in a sea of algorithmic content.

The fear isn’t unfounded: if AI can generate endless music at near-zero cost, why pay human composers—especially for routine work like background scores or jingles? Some foresee AI taking over stock libraries, game soundtracks, and mood playlists, pushing human creators out. Even pop stars face the threat of AI “twins” releasing music in their style faster and cheaper. As the tech advances, these once-hypothetical risks feel increasingly real, fueling urgent calls for guardrails from both industry and government.

Can Policy Catch Up to Creativity?

As AI-generated music accelerates, lawmakers are scrambling to modernize outdated frameworks. In the U.S., Congress introduced the bipartisan “NO FAKES Act” in 2024, which would create a federal right of publicity and outlaw unauthorized digital replicas of a person’s voice or likeness for commercial use. Currently, such rights are covered only by inconsistent state laws. The bill—backed by the Recording Academy, artists’ groups, and major labels—would effectively make AI-generated impersonations like “Fake Drake” illegal without consent.

Another bill, the “Generative AI Copyright Notice and Consent Act,” targets the training process. It would require AI developers to disclose which copyrighted works they’ve used and potentially seek permission via opt-in licensing. AI firms argue such rules would stifle innovation and should fall under fair use. But creators and the U.S. Copyright Office counter that AI’s ability to generate market-ready content makes it materially different from human learning. In a recent report, the Office expressed skepticism that unlicensed training qualifies as fair use when the output competes with the originals.

Europe is moving faster. The EU’s sweeping “AI Act” is expected to mandate labeling of AI-generated content and may require disclosure of training data. Public demand is clear: 89% of Europeans want AI-made music labeled. Member states like France and Germany are pushing for stricter rules, with licensing bodies warning that generative AI could slash creators’ revenues by 27%—nearly €1 billion in those two countries alone.

Elsewhere, Japan and Singapore have taken permissive stances, allowing broad AI training exemptions as long as they don’t “unreasonably prejudice” rights holders. China supports aggressive AI development but mandates labeling and restricts misleading deepfakes. Tencent Music, for instance, has created over 1,000 songs with AI singers—one of which reached 100 million streams.

Whether through laws, licensing systems, or industry standards, 2024 and 2025 have marked a turning point. The future of AI music hinges on whether policy can rein in exploitation while enabling innovation. The stakes are economic, cultural, and deeply personal for those whose creativity is on the line.

An Unstoppable Transformation

Standing at this inflection point, it’s clear that AI-generated music is not a passing fad but a permanent new player on the stage. The technology is advancing at breakneck speed.

Market projections now forecast that the generative-AI music segment will grow from around US $1.7 billion in 2025 to between US $3 billion and US $3.6 billion by 2030. Some analyses forecast even more explosive growth, seeing AI music as part of a broader “AI in music” market that could approach $40 billion by 2033 when including AI-driven music tools, personalized playlists, and more.

According to a report, roughly 60% of musicians already use AI in some part of their process, and most listeners can’t tell the difference: one survey found 82% couldn’t distinguish between AI- and human-made music. For many, if the song hits, it doesn’t matter who—or what—made it.

“A balance needs to be found between protecting artists and encouraging people to experiment.”
HOLLY HERNDON

The challenge now is guiding AI’s rise in music to support, not replace, human creativity. That means new business models—like artists licensing their voices or labels curating AI content—alongside legal protections to ensure fair use and compensation. It also demands a cultural shift, where audiences learn to distinguish and value what’s human-made. As artist Holly Herndon said, vocal deepfakes are here to stay, so “a balance needs to be found between protecting artists and encouraging people to experiment.” Striking that balance won’t be easy.

But even as the cultural conversation tries to catch up, the realities are straining the system. Justin Blau, a DJ/producer and founder of Royal, believes the industry must rethink the concept of copyright itself or risk undermining the democratization of creative accesss.

“The supply of great music will approach infinity, and thus, pro-rata payouts and copyright enforceability will both approach zero,” Blau tells Now Media.

In this new reality, value isn’t defined by scarcity or technical skill. “All of these future songs will have the potential of mass appeal,” Blau says. “As a result, copyright laws and music rights will undergo an irreversible dilution problem.”

“The supply of great music will approach infinity, and thus, pro-rata payouts and copyright enforceability will both approach zero.”
JUSTIN BLAU

Blau’s point is rooted in a broader thesis: when the friction to create music approaches zero, the very meaning of ownership begins to erode. “Recorded music and the idea of skill have nothing to do with each other anymore,” he says. In this world, creation becomes curation, and value shifts away from technical execution toward narrative, identity, and provenance. “Any song that could exist will exist—with either a human or a machine (or both) to thank.”

Against that backdrop, the trajectory feels inescapable. AI will increasingly be a co-author of the songs of tomorrow, whether as a behind-the-scenes assistant or a spotlighted creator in its own right. The industry, long slow to adapt, now grasps the stakes. In the near future, Billboard charts may feature songs written or performed by AI, or by human–AI duos. The question is whether this future will enrich the art form or drown it in soulless mimicry.

For a business built on emotion and originality, the task is clear: embrace AI without losing the human spark. As one label exec put it, “If we had waited for the tech companies to finish innovating [without us], we would have ended up as roadkill.”

Ready or not, the ride has begun, a radical reorientation of music’s future with algorithms in the engine, and artists still (hopefully) at the wheel. The destination is uncertain, but one thing is clear: the music industry as we know it will never be the same.