Distinguishing between music generated by artificial intelligence and a piece composed by a human is becoming an almost impossible listening challenge. This is the finding of an Ipsos study conducted for the streaming platform Deezer, published in October 2025. Conducted among 9,000 people in eight countries (the United States, Canada, Brazil, the United Kingdom, France, the Netherlands, Germany, and Japan), the survey indicates that 97% of listeners were unable to recognize the tracks composed by AI during a blind test featuring two AI-generated tracks and one human-composed track. These results reflect an unprecedented shift: artificial intelligence is no longer content to merely imitate musical structures; it can now reproduce the emotional dimension of sound to the point of fooling nearly the entire audience.
\n\n\n\nWhile the majority of participants acknowledge the technical quality of the generated tracks, their opinions on the artistic legitimacy of these works remain divided. Fifty-one percent believe that AI-generated music will result in songs that are “of lower quality and more generic,” while 64% fear a “loss of creativity” in music production. Nearly half of respondents (49%) nevertheless believe that artificial intelligence can help them discover new works or broaden their musical horizons. These figures reflect a paradox: the public is both fascinated by the technical feat and concerned about the standardization it brings about. Creativity, often seen as the exclusive domain of humans, now appears to face competition from the combinatorial power of algorithms.
\n\n\n\nDeezer, which commissioned the study, is already observing this shift on its platform. In January 2025, 10% of the tracks uploaded daily were entirely generated by artificial intelligence. Ten months later, that figure had risen to 34%—nearly 40,000 tracks per day—according to the company’s internal data. While these tracks still account for a small share of total streams, their growth is exponential. The sudden popularity of the band The Velvet Sundown—entirely created by AI, with one song surpassing 3 million streams on Spotify—illustrates this rapid cultural shift. Music AI is no longer a technological curiosity: it is becoming a massive production force, capable of fueling the industry on a large scale.
\n\n\n\nIn response to this rapid growth, streaming platforms are seeking to balance innovation with responsibility. Deezer has distinguished itself as the first global platform to explicitly label AI-generated tracks with a visible notation on each relevant title. This initiative addresses a growing demand for content traceability and ethical creation. Alexis Lanternier, CEO of Deezer, stated that these results “clearly show that listeners want to know whether they are listening to a track created by a human or by AI.” For its part, Spotify, long accused of a lack of transparency, announced in September 2025 the implementation of new transparency rules encouraging artists and publishers to disclose the use of artificial intelligence technologies in the composition process.
\n\n\n\nThe rise of AI-generated music raises new questions about transparency, intellectual property, and artistic responsibility. While Deezer and Spotify are beginning to label content produced by algorithms, the lack of a globally harmonized framework leaves a gray area regarding the authorship of works and copyright. Should credit be given to the model’s developer, the user who provided the prompt, or the AI itself?
Beyond legal considerations, there is also the question of emotional traceability: when a song resonates with the audience, can we still speak of human expression? By mimicking human sensitivity, AI risks diluting the notion of authenticity that underpins the artistic experience. Aware of these potential pitfalls, several institutions, including UNESCO and the European Commission, are calling for the establishment of transparency indicators and ethical protocols to ensure the integrity of AI-assisted cultural creations1.
These findings spark a profound debate about human perception of sound and creativity. If the ear can no longer distinguish art from algorithms, what becomes of the concept of authorship? Researchers in musical cognition observe that perceived emotion now depends more on the sound structure and listening context than on the composer’s identity. A study by MusicTech Insights (2025) also shows that 72% of young listeners (ages 18–34) believe that a song “remains authentic” as long as it evokes emotion, regardless of whether a human or artificial intelligence was behind its creation2. Musical AI is thus establishing itself not only as a production tool but as a new aesthetic and perceptual language, capable of redefining our relationship with music.
\n\n\n\nBeyond the concerns, the fusion of art and algorithms could open the door to a new form of co-creation. Tools like Suno, Udio, and Mubert already allow artists to compose based on an idea, a text, or a simple vocal tone. According to MIDiA Research (2025), 41% of independent musicians now use an AI generator to speed up their composition process or explore new styles3. This hybridization could redefine the role of the musician, no longer as a mere creator, but as the conductor of a shared musical intelligence, where the machine becomes a partner rather than a substitute.
\n\n\n\nThe Ipsos–Deezer study examines our ability to distinguish between humans and algorithms in music creation. To further explore the relationship between perception and artificial intelligence, read: Vibe hacking: when users manipulate the behavior of generative AI
\n\n\n\n1. UNESCO. (2025). Ethics of Artificial Intelligence in the Creative Industries.
\n https://unesdoc.unesco.org/ark:/48223/pf0000387923 \n
2. MusicTech Insights. (2025). Perception and Emotion in AI-Generated Music.
https://musictechinsights.com/ai-music-emotion-report-2025 \n
3. MIDiA Research. (2025). AI Tools in Independent Music Creation.
https://www.midiaresearch.com/reports/ai-in-music-creation-2025 \n
Distinguishing between music generated by artificial intelligence and a piece composed by a human is becoming an almost impossible listening challenge. This is the finding of an Ipsos study conducted for the streaming platform Deezer, published in…
\n","protected":false},"author":2,"featured_media":440679,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"content-type":"","footnotes":""},"categories":[29],"tags":[59],"class_list":{"0":"post-440676","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-etudes","8":"tag-parlonsia"},"yoast_head":"\n