You have to work with it, not to be eaten by it
Along the Atlantic coast of Cape Verde, musicians, lawyers, and policymakers gathered to wrestle with a question as old as every new tool humanity has fashioned: who controls the instrument, and who does it serve? The African music industry, rich in creative vitality but often underserved by legal infrastructure, finds itself at a crossroads where artificial intelligence offers both liberation — affordable pathways to global stages — and exposure, as weak intellectual property frameworks leave artists vulnerable to synthetic imitation. What emerged from these conversations was not fear, but a measured insistence that the human voice, with all its particular geography and feeling, cannot be averaged away by any algorithm.
- When an unauthorized AI choir remixed a Nigerian artist's song and spread it virally overnight, it revealed how quickly machine learning can outpace both legal protection and an artist's ability to defend their own work.
- Across Africa, patchy intellectual property enforcement means deepfakes and AI-generated plagiarism aren't hypothetical dangers — they are already eroding artists' reputations and market value in real time.
- South Africa's embarrassing withdrawal of an AI policy document riddled with fabricated citations made plain that governments are struggling to regulate a technology that moves faster than institutional understanding.
- At the Atlantic Music Expo, a pragmatic counter-vision took shape: locally governed AI tools, trained on regional music and controlled by the artists themselves, could amplify distinctive sounds rather than dissolve them into global mediocrity.
- The gathering's human heartbeat — a Brazilian seven-woman ensemble performing abroad for the first time, a Bissau-Guinean veteran, a Cape Verdean minister fighting for six million dollars of cultural survival — kept returning to the same boundary: AI can sharpen a tool, but it cannot hold the feeling.
When Nigerian artist Fave discovered an unauthorized AI choir had remixed her song and spread it across the internet before she could respond, she made a calculated choice: she recorded her own version incorporating the AI elements and released it officially. It was a small act of reclamation in a much larger struggle over who owns music in the age of machine learning.
Entertainment lawyer Oyinkansola Fawehinmi read Fave's move as shrewd instinct, but the incident pointed to something more systemic. Across Africa, where intellectual property laws remain inconsistent and enforcement weaker still, artists face genuine vulnerability — not just to plagiarism, but to deepfakes convincing enough to damage reputations and dilute market value. South Africa's government experienced its own humbling lesson when it withdrew a draft AI policy after discovering the document itself contained AI-generated citations, a moment that crystallized the continent's central dilemma: how do you govern something that moves faster than your ability to understand it?
The Atlantic Music Expo in Cape Verde brought that question from abstract to practical. Cape Verde, one of the few African nations with a dedicated AI policy, hosted artists, technologists, entrepreneurs, and officials in conversation about harnessing AI without surrendering creative identity. Culture minister Augustus Jorge de Albuquerque Veiga, working with a budget of just six million dollars, framed AI not as a threat but as a tool to be mastered — lobbying for tourism tax redirections into arts funding and creating diaspora bonds to connect Cape Verdean musicians with communities scattered across Boston and Lisbon. 'You have to work with it, not to be eaten by it,' he said.
What the expo produced was a pragmatic consensus: AI tools for mixing, mastering, and data-driven marketing could level a playing field long tilted toward artists in wealthy countries with expensive studios. José Moura, co-founder of AI startup Sona, argued the critical difference lay in local governance — his platform is built on regional music and controlled by local artists, amplifying what makes a sound distinctive rather than smoothing it into generic competence. 'Artists decide what gets preserved before the AI touches anything,' he said.
The human dimension anchored every conversation. Sambaiana, a seven-woman Brazilian ensemble performing outside their country for the first time, found unexpected kinship in Cape Verde's energy and architecture. Their cavaquinho player was direct: no technology can substitute the feeling that comes from playing, singing, and speaking from lived experience. Veteran singer Patche di Rima offered a quieter truth on the expo's final day — an artist without connection is nothing. AI might serve as media, as a networking instrument, but it could never be the artist holding the room.
In July, the Nigerian artist Fave discovered an unauthorized AI choir had remixed one of her songs, and the synthetic version spread across the internet faster than she could respond. Rather than fight it, she made a calculated move: she recorded her own version that incorporated the AI elements, then released it officially under her name. It was a small act of reclamation in a much larger conversation about who owns music in an age of machine learning.
Oyinkansola Fawehinmi, an entertainment lawyer based in Lagos, saw what Fave did as shrewd business instinct—a way of taking control of a narrative that had already escaped her. But the incident also exposed something deeper: across Africa, where intellectual property laws remain patchy and enforcement weaker still, artists face real vulnerability to having their work copied, remixed, and redistributed by AI systems trained on their own catalogs. The threat extends beyond simple plagiarism into the territory of deepfakes, synthetic versions so convincing they can damage an artist's reputation or dilute their market value.
South Africa's government learned this lesson the hard way. In April, the country withdrew its draft national AI policy after discovering that the document itself contained AI-generated citations—a humbling irony that underscored how quickly the technology had outpaced institutional readiness. The incident crystallized a question that has become urgent across the continent: how do you regulate something that moves faster than your ability to understand it?
At the Atlantic Music Expo in Cape Verde this month, that question moved from abstract to practical. Cape Verde, one of the few African nations with a dedicated AI policy, hosted conversations between artists, technologists, entrepreneurs, and government officials about how the continent's music industry could harness AI's genuine benefits without surrendering its creative soul. Benito Lopes, who has directed the expo since 2024, framed the goal plainly: give performers the knowledge to use these tools "without losing their human identity and their creativity."
Augustus Jorge de Albuquerque Veiga, Cape Verde's culture minister, brought a different urgency to the conversation. With a culture budget of just six million dollars—less than one percent of the national budget—he is fighting to keep local musicians economically viable. He has been lobbying to redirect money from tourism taxes into arts funding and has created diaspora bonds aimed at the hundreds of thousands of Cape Verdeans scattered across Boston, Lisbon, and other diaspora hubs. For Veiga, AI is not a threat to be managed but a tool to be mastered. "You have to work with it, not to be eaten by it," he told the Guardian. "AI will never cover what's authentic. AI is the present already, so we have to discuss this and find ways to work with AI for the country, for the culture and for the future."
What emerged from the expo was a pragmatic consensus: AI-driven tools for mixing, mastering, and data-driven marketing could level a playing field that has always tilted toward artists in wealthy countries with access to expensive studios and marketing budgets. An independent artist in Lagos or Dakar could now use these tools to produce work that competes globally without the overhead that once made such ambition impossible. José Moura, co-founder of Sona, an AI startup that helps artists refine songs through text prompts, argued that the key was local governance. "Homogenisation happens when the tool doesn't know where you're from," he said. Unlike conventional AI trained on global averages, Sona is built on local music and governed by local artists, meaning the technology amplifies what makes a sound distinctive rather than smoothing it into generic competence. "It's the opposite of erasure," Moura said. "Artists decide what gets preserved before the AI touches anything."
The human dimension remained central to every conversation. Sambaiana, a seven-woman ensemble from Brazil performing outside their country for the first time, brought that dimension into sharp relief. Ju Moraes, the lead singer, spoke of recognizing themselves in Cape Verde's energy and architecture, finding kinship across the Atlantic. Her bandmate Rayra Mayara, who plays the cavaquinho, was blunt about what technology could and could not do: "No technology can substitute the feeling we give when we play, sing and talk about our daily lives. AI can complement the production process but it is not a substitute to the human." The veteran Bissau-Guinean singer Patche di Rima, performing on the expo's final day, offered a simpler truth: "An artist without media and networking is nothing." The implication was clear—AI could be media, could be a networking tool, but it could never be the artist.
Notable Quotes
She essentially reclaimed the 'AI version' and released it as her own official expression.— Oyinkansola Fawehinmi, Lagos-based entertainment lawyer, on Fave's response
AI will never cover what's authentic. AI is the present already, so we have to discuss this and find ways to work with AI for the country, for the culture and for the future.— Augusto Jorge de Albuquerque Veiga, Cape Verde's culture minister
No technology can substitute the feeling we give when we play, sing and talk about our daily lives. AI can complement the production process but it is not a substitute to the human.— Rayra Mayara, vocalist with Sambaiana
The Hearth Conversation Another angle on the story
Why did Fave's decision to remix the AI version of her own song matter so much?
Because she refused to be a victim of the technology. Instead of fighting the AI version or ignoring it, she claimed it as her own official work. It was a way of saying: this is my story to tell, not the algorithm's.
But doesn't that just legitimize the unauthorized version in the first place?
Maybe. But in a world where weak copyright laws can't stop unauthorized remixes anyway, Fave's move was about control—getting ahead of the narrative rather than chasing it. She turned a threat into a business decision.
The article mentions Cape Verde has a dedicated AI policy. What makes that unusual in Africa?
Most African countries are still figuring out basic AI governance. Cape Verde has actually thought through what it wants AI to do for its culture and economy. That's rare and deliberate.
José Moura says Sona is built on local music, not global averages. What's the difference?
Global AI learns from everything everywhere, which means it smooths out what makes music distinctive. Local training means the tool understands the specific textures and traditions of a place, so it amplifies uniqueness instead of erasing it.
Is the culture minister's budget of six million dollars realistic for supporting an entire music industry?
It's a starting point, but it shows the scale of the problem. He's having to get creative—diaspora bonds, tourism taxes—because the official allocation is too small. The real question is whether those workarounds can actually move the needle.
What did Sambaiana's presence at the expo signal?
That this conversation isn't just African. It's about artists from the global south—Brazil, Guinea-Bissau, Cape Verde—finding common ground in how they use technology without letting it use them. The energy they felt in Cape Verde was recognition of shared struggle.