Artificial Intelligence (AI) has moved well beyond the point where the music industry could pretend it was simply another piece of software musicians might experiment with.
It is now producing songs at a scale that would have been difficult to imagine a few years ago, turning up in the production process of established artists and, in some cases, creating artists that do not actually exist.
In June, Deezer, the global music streaming service, received around 90,000 fully AI-generated tracks a day. At its peak that month, those tracks made up more than 50% of all new music delivered to the platform.
The figure sounds staggering, although the listening figures tell a different story. Fully AI-generated music still accounted for only between 1% and 3% of total streams on Deezer.
The technology is clearly flooding the industry. The more difficult question is what happens when that technology moves from helping musicians make music to making significant parts of the music for them.
Tyga's latest project, “$TARFACE”, is a useful place to start.
The rapper created a new alter ego inspired by 1980s music and film, drawing on his late mother's love of the era as well as influences including “Scarface”, Prince and Michael Jackson.
The project is deliberately different from the music most people associate with Tyga, with a more cinematic sound built around synth-pop, funk and R&B. He has described it as a full creative departure after more than 15 years in the industry.
Then he confirmed that AI was part of the process.
Tyga said he used AI to create elements including 1980s-style synths and guitar solos, while insisting that the lyrics and vocals were written and performed by him. He also compared the technology with Auto-Tune, arguing that music has always changed alongside the tools used to make it.
That argument is worth considering. Music has never been untouched by technology. Synthesizers changed pop music.
Drum machines reshaped electronic music and hip-hop. Sampling gave producers a way to take something that already existed and turn it into something new. Auto-Tune became so common that it went from being controversial to becoming part of the sound of entire genres.
There is, however, a difference between using technology to alter something you have created and asking technology to create the thing itself.
A guitarist can use a pedal to change the sound of a guitar. A producer can use software to manipulate a vocal.
AI can generate the guitar part, the vocal and potentially the song built around them. That makes the question of authorship much harder to answer.
Doja Cat had her own answer to Tyga's choice.
The musician called him a “penis” for making an AI album during a livestream. The two have previously worked together on songs including “Juicy” and “Freaky Deaky”.
Tyga later defended his use of AI, saying it had been limited to production elements while he remained responsible for the lyrics and vocals.
The reaction might sound like celebrity gossip, but the argument underneath it is much bigger than two musicians insulting each other online.
The music industry has already seen what happens when AI begins to imitate the artist rather than simply assist one.
In 2023, “Heart on My Sleeve” went viral because it sounded as though Drake and The Weeknd had recorded it. Neither musician had.
The anonymous producer Ghostwriter used AI-generated vocals to imitate the two artists and the track amassed millions of views before being removed from major streaming services following objections from Universal Music Group.
The supposed 1960s-style folk-rock band, The Velvet Sundown, took the idea even further. They released music that climbed Spotify's Viral 50 charts in several countries and passed one million monthly listeners.
There were no verified members behind the band in the conventional sense. The project later confirmed that its music was created using AI with human creative direction.
That example matters because the listener was not necessarily looking for AI music. The songs simply appeared in the places people were already discovering music.
The sheer volume of AI-generated music also creates an economic problem. Deezer found that up to 85% of streams on fully AI-generated tracks in 2025 were fraudulent and were subsequently demonetised.
Its concern is not simply that synthetic songs exist, but that they can be produced in huge numbers and used to manipulate streaming systems and divert royalties.
Spotify has faced the same problem. The platform said it removed more than 75 million spam tracks over a 12-month period and has been working on stronger protections against impersonation, fraud and mass-produced music. It is also moving towards clearer labelling of AI-generated artist profiles.
There is clearly an appetite within the industry to experiment with the technology too.
Timbaland has embraced AI openly, working with Suno and developing an AI-focused music venture called Stage Zero. His project includes TaTa, an AI-generated artist, and a new concept he calls “A-Pop”.
Rather than treating AI as something that should be hidden, Timbaland has positioned it as another creative tool and a possible new direction for pop music.
Kanye West's situation is more complicated. A recent lawsuit alleges that AI-generated vocals were used on “Vultures 2” and “Bully”, including claims by a producer who says he created AI voice models and hundreds of vocal generations.
These are allegations and have not been established as fact, but the dispute has added to the growing uncertainty about how much of a recording comes from the artist whose name appears on it.
The money behind this shift is difficult to ignore. A CISAC study estimates that generative AI music could reach an annual value of €16 billion (R299.8bn) by 2028. Under current conditions, the study estimates that 24% of music creators' revenues could be at risk.
That makes this more than a debate about whether Tyga should have used AI to get an '80s guitar sound.
For an artist who cannot afford a producer or a recording studio, AI can lower the cost of making music dramatically. That is a legitimate benefit. It may allow more people to turn an idea into something they can actually hear.
For working musicians, producers, songwriters and session players, the same development raises a harder question. If a machine can do the job, who gets the job?
This is where I find the Tyga argument difficult to ignore, particularly because “$TARFACE” was presented as such a personal reinvention.
Pitchfork rated the album 0.0 criticising its artificiality, and arguing that its retro influences came across as a surface-level imitation rather than a convincing transformation.
Personally, I have no issue with musicians using technology. I would probably be less interested in music if nobody had ever experimented with synthesizers, sampling or electronic production.
There is also something wonderfully messy about human music that technology cannot easily manufacture: the cracked note, the strange breath, the guitar take that was almost thrown away, the producer who spends hours trying to make one imperfect sound work.
Those details tell you somebody was there.
AI can reproduce an era. It can imitate a voice. It can produce a believable guitar solo and, increasingly, an entire song.
What it cannot automatically provide is the life that made the musician want to write the song in the first place.
The music industry does not need to choose between technology and human creativity. It does, however, need to be honest about the difference.
Listeners deserve to know when they are hearing a musician using AI as a tool, when they are hearing an AI imitation of an artist and when the “artist” themselves is a fictional creation.
At some point, the question stops being whether the song sounds good. It becomes whether we still care who made it.
bernelee.vollmer@nationalmg.co.za