Here's What African Musicians Are Saying About AI Right Now

Diplo shared his views on AI-generated music during an interview on the Behind The Wall podcast with Daniel Wall. In viral clips circulating online, he stated categorically that musicians who push back against AI are wasting their time.

"You're not going to win. Like there's no there's no like fighting AI," he said. "You have to just work your best to be the best at it right now. You're wasting your time."

The deejay and one half of Major Lazer isn't exactly the most level-headed voice in the room, and has a troubling history of sexual misconduct allegations, including a recently settled multi-year legal dispute with a woman who accused him of grooming her as a teenager and distributing intimate images without consent. Still, his importance in the global music industry is significant.

OkayAfrica has spent the last few months tracking developments in AI-generated music. We've examined how an AI-generated cover of Stromae's "Papaoutai" racked up millions of streams on Spotify with little disclosure, interrogated what algorithmic listening means for African artists in an era when emotional authenticity has become optional, and profiled Rwandan artist Elvin Cena, whose AI-assisted "Let Me Be" became an unexpected viral hit after he turned to Suno to rework a song he had shelved. As the music embeds itself in our lives, it remains important to track these changes; to commit to memory how they intersect with issues of ownership, labor, and the extractive practices that power global capital, and to record a multiplicity of African voices on the registry.

Live musicians aren't on the margins of this discussion. In fact, they're at the core of it, since they make the music that AI engines are trained on. Rather than ushering in the era of artificial general intelligence that its most vocal proponents promised, the rise of generative AI has only deepened socio-economic and environmental issues, and the question of labor and how it's compensated fits squarely into that framework.

Navigating AI: Concerns Over Education, Labor, and Copyright

Benjamin Jephta — the South African bassist and composer whose landmark debut Homecoming shook the country's improvised music scene a decade ago, and who returned last year with the sprawling Homecoming Revisited — approaches the conversation around generative AI with the same careful, considered ear he brings to his arrangements.

"What is striking to me about generative AI in music is that the more I encounter it, the more questions I have," he says to OkayAfrica.

For now, he uses AI mostly as a productivity tool rather than something embedded in his creative practice. He was recently part of SAMTIC, the South African Music Technology, Innovation and Capacity-building Project, an Erasmus+ initiative linking universities in South Africa and Europe on digital music education and training, including AI-driven production. The work he found most compelling was AI being used in live settings alongside real musicians rather than replacing them.

"That kind of use felt more compelling to me because it at least tried to imagine AI in relation to human performance, responsiveness, and collaboration, instead of treating music-making as something that can be fully automated."

But the trip also confirmed his deeper concern: there is little serious thought or strategy guiding generative AI's entry into music education. The default response, he notes, tends to be that students are going to use it anyway, so institutions may as well incorporate it, which is not the same as having a pedagogical or ethical framework. Even Berklee Online now teaches courses on AI in music and AI for songwriters, a sign of how quickly the technology is being normalized.

Jephta's caution sits inside a recurring tension that surfaces whenever something incubated in the margins starts moving into the mainstream. The moment an underground artist breaks through, resistance often follows - talks about legitimacy abound. The work stops being received on its own terms and gets filtered through expectations that demand it remain authentic in some narrow sense, or else justify its popularity on the spot. 

That shift flattens the complexity of what made the work resonate in the first place, and the transition into visibility tends to trigger a kind of interpretive panic, where suddenly everyone is policing what something is supposed to be.

The same pattern is increasingly evident in how AI is being received now that it has entered mass consciousness. It has become impossible to treat it as niche infrastructure or speculative tech, and the conversation has subsequently shifted into a more emotional register, swinging between hype and alarm. 

The concrete harms are real and need to be accounted for: labor displacement, authorship questions, data extraction, and the flattening of creative work. Alongside those, new systems tend to arrive with disruption, then get metabolized into everyday practice. Panic shapes what people are able to see, and what they refuse to recognize as already integrated into the present.

As The Brother Moves On's Siya Mthembu puts it, speaking to this broader cultural pattern of technological and aesthetic panic: [insert quote here]. That kind of framing helps name the emotional texture of the moment without collapsing it into either blind acceptance or total refusal.

"I think new technologies make humans very nervous," says the artist, who also has a solo project as Hymn Self. "The commercial sector throws in this idea of 'you're gonna miss a moment, it's too fast! [We saw it] with DVDs, with Blu-Ray, and sometimes [it was done] so much that people got over the thing. I think that's where we're going with AI; because they're doing it so much, they're damaging the utility, and that's the danger of it. That means certain people won't use it. My nine-year-old is afraid of AI, because everything that she watches on TV is telling her that AI is gonna turn into The Terminator. Us who grew up on [that] will tell you that there's nothing wrong with machines. I don't think AI is gonna take anything that human beings won't be able to make."

Johannesburg-based artist Itai Hakim sharpens the legal stakes. "For a South African musician, the law protects the song but not the singer," he notes. "The Copyright Act of 1978 protects the fixed recording. It does not protect your voice frequency, your vocal style, or the unique timbre that makes you recognizable." 

The Protection of Personal Information Act classifies the voice as biometric data, but carves out an exception for anything "made public," and every Spotify upload qualifies.

The platforms, he argues, have inverted the logic of Creative Commons. What was designed to let artists say "take my work, but not my credit, not my commercial value" now reads, buried in the terms of service, as: "you gave us your work; we'll train our AI on it; and you will receive nothing." The legislation meant to address any of this is stalled. 

Hakim points to the Braille case: in 2022, the Constitutional Court gave Parliament 24 months to fix the Copyright Act. The President has twice refused to sign the Bill. It was first introduced nine years ago. The AI Policy's comment period closes in 50 days.