Thirty-eight years ago, Neneh Cherry crossed her arms, cocked her head and dared the world to look away. Now half the songs hitting streaming services were made by software that has never had attitude in its life. Somebody get this machine a bomber jacket.
Arms Crossed, Chin Up, 1988
Let’s set the scene. It’s late 1988 and British pop is a conveyor belt. Stock Aitken Waterman are cranking out hits with the efficiency of a Toyota plant, every chorus buffed to the same high gloss, every singer interchangeable enough that you could swap Rick for Kylie and only their mums would notice. Into this beige wonderland strolls a Swedish-born, London-bred, jazz-raised former punk named Neneh Cherry, and she does not ask permission.
“Buffalo Stance” was a mongrel in the best possible sense. Hip-hop drums, pop synths, a rap flow that sounded like it had been picked up on a Ladbroke Grove street corner, and a hook built for the dancefloor. It started life years earlier as a B-side tucked behind Morgan-McVey’s “Looking Good Diving,” got pulled apart and stitched back together with help from Bristol’s Wild Bunch orbit and producer Tim Simenon of Bomb the Bass, and emerged as something nobody had a box for. It went Top 3 on both sides of the Atlantic, landed on the Slaves of New York soundtrack, and opened her 1989 debut Raw Like Sushi.
The title itself was a wink to people in the know. “Buffalo” was the name of stylist Ray Petri’s London crew — the boys in MA-1 jackets, cycling shorts and pork-pie hats who were all over the pages of The Face and i-D. The stance was the pose: arms folded, head turned, a look that said I’ve already decided what I think of you. Cherry took a fashion-mag attitude and turned it into a three-and-a-half-minute manifesto about not getting played — by men, by money, by anybody.
And then there was Top of the Pops. Cherry performed the song heavily pregnant, in a bodycon outfit, bouncing around like the BBC hadn’t been clutching its pearls since 1964. Her mother had told her not to wall off motherhood from the rest of her life, and she took that advice live on national television. In one appearance she torched about four separate rules for how women in pop were supposed to behave.
The Original Synthetic Scandal
Here’s the delicious bit of timing. The same year “Buffalo Stance” was climbing the American charts, Milli Vanilli were selling millions of records with voices that weren’t theirs. When the backing track jammed at a 1989 concert and the truth came tumbling out the following year, they became the first act to have a Grammy taken back. The outrage was biblical. People returned albums. Lawyers got rich. America decided, collectively and loudly, that it cared very much who was actually singing.
Hold that thought, because it’s about to age like milk.
Enter the Machines
Fast-forward to April 2023, when an anonymous producer calling himself Ghostwriter dropped a fake Drake-and-The-Weeknd track that sounded close enough to the real thing to send label lawyers into cardiac arrest. Fourteen months later, in June 2024, the major labels sued AI music generators Suno and Udio, accusing them of training on their catalogs without asking. The startups argued fair use. The labels argued theft. Everyone hired more lawyers.
Then, in the summer of 2025, a band called The Velvet Sundown racked up streams on Spotify before the internet figured out there was no band — just a well-prompted algorithm and a stock-photo aesthetic. By autumn, Xania Monet, an R&B persona built with AI by a human poet named Telisha Jones, had landed on Billboard’s R&B Digital Song Sales chart, become the first AI act to earn enough radio play to crack a Billboard airplay chart, and reportedly signed a multimillion-dollar deal with Hallwood Media after a bidding war. In November 2025, a rendered cowboy called Breaking Rust topped Billboard’s Country Digital Song Sales chart for two weeks. Somewhere, a Nashville songwriter threw his Stetson into a lake.
Remember that biblical Milli Vanilli outrage? The modern equivalent was a shrug and a playlist add.
The Labels Swap Their Pitchforks for Pens
The industry’s response went from “see you in court” to “let’s do lunch” with breathtaking speed. Universal settled with Udio in October 2025 and announced plans for a jointly licensed AI music platform. Warner settled with Suno in late November 2025, with Suno agreeing to build new models trained on licensed music. Warner squared things with Udio too. Sony, meanwhile, kept the gloves on, and by mid-2026 UMG and Sony were still fighting Suno in a Massachusetts courtroom. Upstart Klay became the first AI music company to get all three majors on board, pitching itself as the licensing-first good kid in class.
Translation: the people who sued the robots for stealing music have now decided to sell the robots the music. That’s not a moral awakening. That’s a stance — and it’s very much a Buffalo one, arms folded and eyes on the money.
The Flood
The numbers this year are frankly unhinged. Deezer reported in April 2026 that 44 percent of new daily uploads to its service were AI-generated. By July, that figure had crossed 50 percent — roughly 90,000 synthetic tracks a day at June’s peak, compared with about 10,000 when the company started detecting them in January 2025. That means more than half of what gets shoveled into the streaming pipeline every morning was written by nobody, performed by nobody, and dreamed up by a text box.
The gatekeepers have started building fences. The EU AI Act’s transparency rules, which include labelling synthetic audio, became enforceable on August 2. The IFPI published global principles on July 30 for how AI-built recordings can qualify for official charts, and Australia’s ARIA went further, declaring wholly AI-generated recordings ineligible starting with the chart dated August 31. Apple Music has reportedly begun bouncing some AI submissions at the door. The velvet rope is back, and this time it has a detection algorithm.
What a Machine Can’t Fake
So here’s the question that should keep every prompt jockey up at night: could Suno have made “Buffalo Stance”?
Technically? Give it a prompt like “late-’80s London hip-hop pop, female rapper, Roland synths, sassy” and it’ll spit out something with the right drum sound and a passable flow in about forty seconds. It might even be catchy. What it can’t do is spend three years dragging a song from a B-side to a hit, soaking up Bristol sound-system culture, Petri’s styling, punk squats and a jazz-trumpeter stepdad’s record collection along the way. It can’t walk onto Top of the Pops seven months pregnant to make a point about women’s bodies and ambition. It can’t mean it.
The Buffalo stance was never a sound. It was a posture — a refusal to be packaged. Which is the exact thing generative AI is built to do: package. It averages. It smooths. It hands you the median of every song it has ever ingested, which is basically what Stock Aitken Waterman were doing in 1988, just with fewer mullets. The irony is thick enough to spread on toast: AI music’s natural enemy isn’t the human musician. It’s the attitude Neneh Cherry was selling in the first place.
And Cherry herself has already shown what the alternative looks like. In 2022 she handed “Buffalo Stance” to Robyn, who slowed it down and stripped it back with producer Dev Hynes, while Mapei took the rap. The video starred Pose‘s Indya Moore and was teased on Trans Day of Visibility. That’s how you remix a classic: not by feeding it into a machine and asking for more of the same, but by passing it to someone who loves it and letting them bend it into something new. Even in March 2026, podcasters were still dissecting how a flipped B-side became an era-defining hit — a reminder that the magic was always in the human who heard something no one else did.
The Stance, Updated
None of this means the robots lose. Half the uploads are already theirs, the labels are cashing licensing checks, and some listeners genuinely don’t care whether their chill R&B was rendered or recorded. Xania Monet’s success, after all, came from a human writer using AI as her instrument — which is a lot closer to Cherry chopping up a B-side with a drum machine than the doom-posters want to admit.
But if you’re a musician staring down the barrel of 90,000 synthetic tracks a day, the 1988 playbook still works. Cross your arms. Turn your head. Make something with a point of view so specific that no algorithm averaging the entire history of recorded sound could ever land on it by accident.
The machines can do the beat. They can do the hook. They can even do the voice. What they still can’t do is the stance.
About The Author
