Generative AI Is Changing the Way We Consume Music by African Artists

You’ve probably seen and heard it by now: a wave of videos circulating the internet, soundtracked by what feels like a reprise of Stromae’s decade-old “Papaoutai.” The arrangement instantly grabs the listener — swelling choirs, orchestral flourishes, war drums engineered for maximum emotional pull. In one clip, a close-up of the artist’s side profile cuts to a wide shot of a sparsely populated room. It looks real, but something is off. It could be AI. It could also simply be the result of platforms like YouTube baking AI upscaling directly into Shorts, making the line between synthetic and human increasingly difficult to detect.

In another widely shared iteration, social media creator Arsene Mukendi delivers a raw, emotive vocal performance over the same instrumental. The clip racked up over a hundred thousand views on Instagram alone before he later revealed that the cover was AI-generated. By then, the song had already done its work.

Titled “Papaoutai (Afro Soul Version),” the track is credited to Chill77 and Unjaps, with copyright and publishing attributed to UnjapsAB. Uploaded on December 21, 2025, it has since amassed over 14 million Spotify streams. Hundreds of creators now use it across platforms, while many more encounter it algorithmically — without context, disclosure, or any indication that the voice they are responding to does not belong to a person at all.

The playbook here is familiar. Across a growing number of AI-generated releases, the strategy is to target the listener’s emotional core: swirling strings, power vocals, dramatic crescendos that taper into restraint. The same UnjapsAB profile previously uploaded an AI-powered reimagining of Sisqó’s “Thong Song,” built on the same recognizable effect and immediate emotional payoff.

At this point, ethical questions become unavoidable. “Papaoutai,” a deeply personal song about Stromae’s father, who passed during the Tutsi genocide in Rwanda, remains the biggest record of the Belgian artist’s career. Its official video has surpassed a billion views on YouTube, and its influence has travelled far beyond Europe. In South Africa alone, the song has been interpolated and referenced repeatedly — from Dr Malinga’s “Oteng” to a recently uploaded DJ Maphorisa Instagram reel featuring a song snippet that nods directly to its lyrics.

It is the result of what happens when platforms quietly present listening as passive consumption, as opposed to encounter and experience. What, then, are listeners being trained to overlook? And what happens to the idea of art when emotional authenticity becomes optional? 

A 2025 survey conducted by Deezer in partnership with the global market research and consulting firm Ipsos offers a revealing data point. Surveying 9,000 respondents across eight countries, the study found that 97% of participants were unable to reliably distinguish between fully AI-generated music and music made by humans. Read superficially, the figure might suggest that concerns around AI authorship are overblown, or that listeners simply do not care.

However, a more troubling interpretation is that the conditions for meaningful choice no longer exist. If listeners cannot tell the difference — and if platforms decline to tell them — then consent becomes impossible. In this context, Deezer’s decision to introduce an AI-tagging system is less a technological innovation than an ethical intervention. It implicitly acknowledges that transparency, not detection accuracy, is the real issue at hand. Notably, Deezer remains the only major streaming platform to implement such a system.

Elsewhere, AI-generated tracks circulate without disclosure, folded seamlessly into recommendation engines that treat human labour and machine output as interchangeable inputs.

Spotify, in particular, has faced criticism for flooding lo-fi and ambient playlists with low-effort, AI-generated tracks, a move that effectively displaced human producers within a genre already operating at the margins. The platform recently fired back, insisting that it does not create or own music, nor promote or penalize tracks created using AI tools. It has also pointed to safeguards it claims are in place to prevent AI impersonation.

Yet “Papaoutai (Afro Soul Version)” sits uneasily within these assurances. The track does not directly use Stromae’s recordings, but it functions as an aesthetic and emotional stand-in for the original, reproducing its melodic contours, dramatic arc, and vocal affect with remarkable fidelity. That it has accumulated millions of streams on a platform that claims to guard against impersonation exposes a central contradiction: Spotify’s policies appear to intervene only when impersonation is explicit, rather than when it is structurally or sonically obvious.

If platforms police voice cloning but allow near-perfect affective mimicry, are they meaningfully protecting artists, or merely protecting themselves legally? This is a question of performative identity. The original lyrics — “dites-nous qui donne naissance aux irresponsables?” (“Tell us who gives birth to irresponsible people?”) — are intimate and accusatory, a cry for a father that Stromae never knew. What does it mean when a song sounds like grief that isn’t yours?

All of this returns us to a music ecosystem that is increasingly unsustainable for artists. If humans are forced to compete with machines for a share of an already diminishing pie — machines capable of producing vastly higher volumes of content — then what does that say about the value of art itself? Do artists, and their catalogues, become little more than training data for generative models to mine?

The recent deals that Universal Music and Warner Music have undertaken would certainly suggest so. The labels came out guns blazing in June 2024, accusing the AI song generator tools Udio and Suno of massive copyright infringement. By the time October 2025 came, Universal Music had settled with Udio, and Warner Music did the same with Suno and Udio. As one writer eloquently put it, the strategy is to “sue for leverage, negotiate for profit, settle for partnership.”

A joint PMP Strategy and CISAC study on the economic impact of AI in the music and audiovisual industries estimates that generative AI outputs in music will be worth a cumulative €40 billion over the next five years, rising to an annual value of €16 billion by 2028. By that point, generative AI music is projected to account for around 20% of streaming platform revenues and nearly 60% of music library revenues.

“Papaoutai (Afro Soul Version)” has entered the Global Spotify chart, yet another marker in an ecosystem slowly normalizing AI as part of our daily lives. AI-generated music can succeed, and this is apparent in the success of artists such as Xania Monet and Sienna Rose. The question: what gets lost when emotion becomes programmable, and authenticity optional?