‘I feel like I’m at war’: are we losing the battle against machine-made music?
This year, the battle for song of the summer has been eclipsed by a much more complicated – some would even say disturbing – debate. That’s because we find ourselves asking not “What’s the song of the summer?” but rather “Is the song of the summer even real?”
Among the top contenders for the title is Fenix Flexin’s Rubberz, a single released in June that has ascended to No 58 on the Billboard Hot 100 and racked up more than 35m Spotify streams. It’s not the sort of fare Fenix usually cooks up. The artist is known for his trap music as part of the rap duo Shoreline Mafia, but Rubberz is a mildly noirish, 80s-inspired synth-pop track featuring a voice nothing like his. The song has drawn comparisons to Morrissey, but it more closely resembles Men at Work’s Down Under, or a Weird Al Yankovic parody of Men at Work. It’s pretty awful. But more importantly, it has an uncanny quality to it. It sounds off.
Almost immediately after it came out, skeptics accused Fenix of employing artificial intelligence to create the track, if not entirely at least partly. He and the song’s credited producer, Purps on the Beat, initially denied it, with Fenix insisting that he’d just used Auto-Tune and a phoney British accent. The kerfuffle culminated this month when indie DJ-producer Medasin posted a video that he said proved that Rubberz had been made using the free generative music tool Treblo.
Medasin then captured himself giving Treblo prompts to pump out his own track. Its sound and lyrics were eerily close to those of Rubberz. Fenix refuted Medasin’s claims, posting what he said were clips of the song’s stems, or its fundamental components, but then Medasin posted a rebuttal of him demonstrating how AI songs, too, can be broken down into stems. A few days later, Fenix finally admitted to using AI to make Rubberz, writing: “never said I didn’t use AI, i said it had nun to do w recording process.”
This is in line with the findings of the platform’s own open-source classifier, Was This Made With Treblo?, with finds Rubberz to be “very likely Treblo” according to Wired. The tool hasn’t been independently reviewed, but Treblo co-founder Ryan Tremblay confirmed to the Guardian that the classifier returns a very high probability score (“higher by far than any false positive”) for Rubberz, indicating that Treblo generated the song. “We built Treblo for everyone, from normal people experimenting at home to charting artists looking for new sounds,” Tremblay said. “Some artists appear to have used our model to find a sound that millions of people connected with, and that’s exciting.” When reached by the Guardian, Fenix’s representative declined to comment.
The uproar over Rubberz coincided with another major development in music in recent weeks. On 29 July, several record companies, including “the big three” – Universal Music Group, Sony Music and Warner Music Group – issued a proposal calling for global music charts to disqualify AI-generated tracks from consideration if they did not meet certain criteria. According to their suggested guidelines, a song could enter the charts only if it was properly authorized, lawful and substantially human-made, and does not raise concerns of streaming or chart manipulation. The issue, as the record companies stated in a press release, is that a significant number of AI-generated tracks hitting streaming services like Spotify have used tools trained on artists’ music without their consent and are often linked to fraudulently boosted streaming numbers. Under the proposed rules, these songs would not be allowed to enter charts such as the Billboard Hot 100.
At the same time, all the major record labels have announced partnerships with the likes of Nvidia, Spotify and Stability AI to develop AI music tools, with BMG announcing a partnership with the AI music-creation platform Suno this week. The rapper Tyga, a labelmate of Fenix, proudly told Vibe magazine recently that his new album $tarface was created with AI, saying: “There’s nothing wrong with using technology as a tool.” (The album received Pitchfork’s first 0.0 review in nearly two decades.)
He is not alone in his permissive attitude. While Rubberz is the first AI-generated single to crack the Hot 100, it is not the first to land on the Billboard charts. In November, the AI act Breaking Rust hit No 1 on the country digital song sales chart with Walk My Walk, a track credited to a person who goes by Aubierre Rivaldo Taylor. That same month, AI-powered R&B singer Xania Monet – whose name instantly conjures actual human R&B singers Victoria Monét and Janelle Monáe – became the first AI act to enter a Billboard radio airplay chart, later climbing to No 1 on the R&B digital song sales chart. An avatar that Mississippi poet Telisha “Nikki” Jones invented with Suno, she reportedly landed a multimillion-dollar record deal last fall.
After Xania Monet’s milestone, Billboard reported that a total of at least six AI-generated or AI-assisted artists had debuted on its various charts in 2025. The publication admitted, however, that the tally could be higher because it has become harder and harder to determine which songs have used AI, and just how much. This was validated a week later when a study by the French streaming platform Deezer and research firm Ipsos revealed that 97% of its participants couldn’t tell the difference between AI-generated music and human-made music.
Unsurprisingly, the proliferation of AI artists has struck a nerve with the human ones. SZA told i-D: “I feel like I’m at war because of AI.” Last month, she slammed AI generators like Suno and alleged that they were being trained on her songs without her permission. “If you’re a musician and you support this degenerate shit? You’re disgusting and there’s nothing you could ever say to me to make this OK,” she said. Jack Antonoff recently called AI-music makers “godless whores.” Billie Eilish, Bad Bunny, Kate Bush, Annie Lennox, Nick Cave, Sting, Stevie Wonder and Paul McCartney have all protested the heavy use of generative AI in music.
And why shouldn’t they? Yes, dictating what one wants to create to an app requires a sliver of inspiration and imagination, but it pales in comparison with the blood, sweat and tears artists pour into music of their own – music that hasn’t been trained, often without permission, on reams of pre-existing copyrighted work. If those artists’ labor is grueling, the results, performed live and perfected in a studio by bands and session players with real instruments, can be deeply moving and totally exhilarating. Directing prompts to AI music generators like Suno, Udio and Treblo to spit out fully formed songs for mass consumption is a faster, cheaper, easier way to make music. It also disrespects and debases the art form, reducing hard-earned craft to half-assed mimicry.
If the outcry against AI sounds like a bunch of intransigent Luddites shaking their fists at the inevitable arrival of a new era, know that naysaying artists and wary record companies are not alone. A recent report published by the music and entertainment insights company Luminate showed that listeners, especially members of gen Z and gen Alpha, are increasingly uneasy with AI-generated songs. The study found that 44% of US respondents felt less inclined to listen to a song if it involves generative AI. “Consumers are net negative,” a Luminate analyst told NPR in May. “People are more likely to feel uncomfortable than to feel comfortable with AI use.”
Universal, Sony and Warner are not demanding an outright ban on AI music from charts and streaming services. Though all three sued the AI music generators Suno and Udio in 2024, alleging they had trained their platforms on the labels’ artists without their permission, Universal settled with Udio and Warner settled with both Suno and Udio last year. (Sony has yet to settle with either.) The labels continue to work with Spotify and AI companies to develop generative music tools, in a move that effectively keeps their options open. Warner executive Robert Kyncl said the deals would foster “new creative and commercial possibilities”, adding: “Now, we are entering the next phase of innovation. The democratization of music creation.” Universal, Sony and Warner did not respond when approached by the Guardian for comment.
Meanwhile, other artists have expressed support of AI as a means to enhance songwriting. “With new technology, I try to use it as a tool to grow, not a crutch,” Björk wrote in a 2025 post on X. “If there is no soul in tomorrow’s music made by AI, it is because no one put it there … Anything that is mass-manufactured without the attention of creativity is that way, AI or not. So it is not about the tool – it is what you do with it.” Even McCartney enlisted AI to isolate John Lennon’s vocals for the final Beatles song, 2023’s Now and Then. What these labels and so many musicians want is a system of guardrails to ensure that AI use is limited, honest and transparent.
There is hope. In January, Bandcamp banned all music fully or substantially created by AI from the platform. Though Tidal allows fully AI-generated songs on its app if they are not linked to fraud or artist impersonation, they cannot collect streaming royalties or direct-to-fan sales revenue, and get tagged with an “AI” label. A coalition of music industry organizations, including the Grammys and the Recording Industry Association of America, recently announced that it was taking steps to notify listeners of AI use in recordings. This month, Spotify said that it would label AI artists with an “AI Persona” tag and exclude these artists from its personalized recommendations, while its optional “AI Credits” feature allows artists to note whether AI was used in a song’s creation.
Per the record companies’ proposed criteria, such AI Personas would not be chart-eligible. If Deezer’s findings are any indication, plenty of listeners would support this. Its 2025 survey results showed that 80% of respondents agreed that entirely AI-generated music should be clearly labeled to listeners, while 52% thought entirely AI-generated songs should not be included alongside human-made songs in the main charts.
Author and journalist CJ Farley echoed this sentiment in a recent column for Time. He dismissed the terms “AI artists” and “AI musicians”, suggesting that we instead refer to AI releases as “AI products”. Then he took it a step further, arguing that these products should be confined to their own chart “away from human musicians with real lives and with real financial obligations”.
If Rubberz is a joke, Fenix Flexin might have been trying to see exactly how much he could get away with. (That would at least explain why the song is such dreck.) But what about all the AI products with more nefarious purposes floating around out there undetected, the slop created by impostors and opportunists attempting to steal the work of others and pass it off as their own? There’s more at stake here than chart rankings. Artists’ livelihoods and legacies are on the line. If we don’t act now, tomorrow’s song of the summer may just be a sad, hollow shell of yesterday’s.