Source: The Conversation – Canada – By Cate Cleo Alexander, Postdoctoral research fellow, University of Toronto
Would it be obvious if artificial intelligence (AI?) created your new favourite song?
Millions of listeners have recently encountered that question through a viral Afro-soul cover of Papaoutai, the 2013 hit by Belgian artist Stromae. The cover has skyrocketed in popularity across streaming platforms and social media.
But unknown to most audiences, it was created using AI, according to Deezer, a French music-streaming service.
The Afro-soul cover highlights a growing challenge — the difficulty identifying when generative AI has been used in production — and how audiences, platforms and artists are struggling to respond.
When Stromae first released the upbeat dance song Papaoutai as part of the album Racine carrée, it topped the charts in Belgium, France, Germany, Israel, the Netherlands and Switzerland. More than a decade after release, it’s still one of the most-viewed French-language songs on YouTube.
Some 12 years later, in December 2025, an Afro-soul cover of Papaoutai was uploaded to Spotify. While it’s hard to track the exact reach of the song due to various removals and re-uploads on YouTube, Instagram and TikTok, the song currently has almost 80 million streams on Spotify.
The authorship of the Afro-soul version is commonly attributed to mikeeysmusic — a Swedish musician with a verifiable social media presence and discography — Chill77, whose identity is difficult to verify, and Unjaps, an independent record label. None of the artists have made a public statement about the controversy.
Why does all of this matter? Most music platforms lack clear labelling for AI music, and this places the difficult task of identification on listeners.
Identifying AI use in music production
AI-generated music has become a very broad category. As machine learning engineer and researcher Christopher Landschoot argues, the term AI-generated music “is casually tossed around whether AI is used to emulate an effect, automatically mix or master, separate stems, or augment timbre. As long as the final audio has been touched in some way by AI, the term gets slapped on the entire piece.”
It’s hard to tell how and to what extent AI was used in the making of the cover of Papaoutai. Did mikeeysmusic and Chill77 upload Stromae’s original song into an AI program, like OpenAI’s Sora, give it a command and upload the AI-generated result? Did they train an AI program on the vocals of another musician, Arsene Mukendi, to generate choir vocals? Or was the cover an iterative process where the artists fine-tuned and edited the output?
Does it matter when the lyrics and melody were written by someone else?
Identifying AI use in music is difficult, even for scholars like us who study generative AI. A study published by Deezer-Ipsos, which surveyed 9,000 people across eight countries, found that 97 per cent of people couldn’t tell the difference between fully AI-generated music and human-authored music.
A big contributor to the confusion is the lack of response from platforms. While Bandcamp has taken a clear anti-AI stance and works to keep AI-generated music off the platform entirely, other platforms like Spotify have gestured towards governance changes but largely allowed AI music to rack up streams without clearly disclosing the use of AI.
The popularity of short-form videos (like on TikTok), in which users encounter uncontextualized song snippets, further propels the prominence of AI-generated music.
As one journalist argued:
“If listeners cannot tell the difference — and if platforms decline to tell them — then consent becomes impossible.”
Emotional responses on social media
In comment sections, audiences are often surprised to learn that the song was created using AI. Many have praised the cover, describing it as “a lot more instrumental, emotional and grand.”
But these positive feelings abruptly shift upon learning about AI use in the song’s creation. As another Reddit user commented:
“I’m actually so sad it [is AI]-generated. It sounds wonderful, but I personally can not support [AI] taking over creative industries such as music and art. And I know there are plenty of African choirs who could have nailed the vision without the use of [AI].”
Whether audiences choose to listen to AI-generated music is often framed as a moral decision. This is complicated, however, when it becomes increasingly difficult to discern what music is AI-generated or how generative AI was used in its creation.
According to the same Deezer-Ipsos study, 73 per cent of people surveyed “think it’s unethical for AI companies to use copyrighted material to generate new music without clear approval from the original artist.”
Stromae has remained quiet on the issue so far, with audiences speculating about his response to what some see as an AI appropriation of a very personal song written about his father, who died in the Rwandan genocide.
Many English-speaking users, unaware of this context, have used the heavy drums of the Afro-soul version to soundtrack everything from fashion haul videos to comedy shorts. As one TikToker asked:
“Did you ever think that the song about losing your father in the Rwandan genocide would be used, in your lifetime, to post gym thirst traps?”
AI as a tool for remixing
Artists have been grappling with the possibilities and concerns of AI use in music production, but there has been a precedent for “playing” with original songs.
Remix — through cassettes, spin tables and synthesizers — has been a part of music fan culture for decades. There is a rich tradition of fans using technology to cut, copy, paste, play, reimagine and recontextualize music.
Can AI-generated covers — the newest way of using technology to “play” with music — be understood as part of this legacy of remix?
On the legal side, all seems above board — at least in France. In the case of the Papaoutai cover, the French Society of Authors, Composers and Music Publishers has upheld its legality. They note that Stromae is “properly credited” and that royalties will be shared between Stromae and the record label that produced the cover.
So is it remix? Maybe. Is it legal? Apparently.
But as seen in this example, audiences still struggle with songs extracted from their original context and created with AI technologies, which are themselves inherently extractive.
Does this context shift the perception of this song as a form of extractive remix production?
The Afro-soul cover of Papaoutai illustrates how quickly AI-generated music can circulate. It also signals an increasing amount of debate as artists, audiences, and platforms navigate the future of AI-generated music.
Cate Cleo Alexander has received funding from SSHRC (Doctoral Canadian Graduate Scholarship) and is a member of the Creative Labour and Critical Futures (CLCF) project.
Lauren Knight receives funding from SSHRC (Doctoral Canadian Graduate Scholarship) and is a member of the Creative Labour and Critical Futures (CLCF) project.
– ref. Could you tell if your favourite song was made with AI? The viral ‘Papaoutai’ cover controversy suggests not – https://theconversation.com/could-you-tell-if-your-favourite-song-was-made-with-ai-the-viral-papaoutai-cover-controversy-suggests-not-274607