Ipsos Poll Reveals 97% of Listeners Cannot Distinguish AI-Generated Music from Human Compositions

A global study commissioned by the streaming service Deezer and conducted by Ipsos has revealed a striking lack of discernment among music listeners regarding the origin of content. The vast majority of listeners are simply unable to differentiate between musical works created by artificial intelligence and those produced by human artists. In a blind testing scenario involving 9,000 respondents across eight nations—the United States, the United Kingdom, France, Brazil, Canada, the Netherlands, Germany, and Japan—a staggering 97% of participants failed to correctly identify the source of three compositions they heard. This outcome was met with surprise by 71% of those polled, and over half (52%) admitted feeling uneasy about their inability to distinguish between machine-generated and human creativity.

The findings underscore a critical need for transparency in the digital music landscape, according to Alexis Lanterne, CEO of Deezer. A significant 80% of participants believe that music entirely generated by AI should carry clear labeling. Furthermore, 73% of respondents deemed it unethical for neural networks to be trained using copyrighted materials without the explicit consent of the artists involved. The sheer volume of AI content necessitates action: Deezer statistics show that approximately 50,000 fully AI-generated tracks are uploaded to the platform every day, representing roughly one-third of all new uploads as of November 2025. In response, Deezer has begun marking such content, excluding it from editorial playlists and algorithmic recommendations to minimize its impact on the royalty pool.

Streaming platforms are taking coordinated steps toward regulation. Spotify, for instance, has updated its policies to mandate labeling, implement spam filtering, and protect against unauthorized voice cloning. The scale of the problem is immense; over the past year, Spotify removed 75 million “spam tracks” that were degrading the user experience and siphoning royalties away from legitimate performers. The urgency of these measures was highlighted by the success of synthetic projects, such as the AI band “The Velvet Sundown,” which amassed one million monthly listeners on Spotify before its non-human origins were disclosed.

Beyond platform policies, the industry is also moving toward legal resolutions and strategic alliances. A major lawsuit involving Universal Music Group (UMG) and the AI music development company Udio reached a settlement in late October 2025, paving the way for market legalization. Initially, UMG, alongside Sony Music and Warner Music Group, had filed suits against both Udio and Suno for training their models on protected intellectual property. As part of the global settlement, the parties announced a joint venture to develop a new subscription platform, slated for launch in 2026. This platform will utilize generative AI trained exclusively on authorized and licensed music, a move that Sir Lucian Grainge, Chairman of UMG, stated demonstrates a commitment to safeguarding artists' interests.

The Ipsos survey also brought consumer anxieties to the fore: 51% of respondents believe that the proliferation of AI will lead to an increase in low-quality music, and nearly two-thirds fear a general decline in creativity across the industry. The inability of listeners to differentiate between AI and human compositions challenges the fundamental value of artistic attribution. However, the decisive actions taken by major labels and streaming platforms signal the formation of a new regulatory framework designed to ensure that the human element remains central to the music experience.