Why listeners can’t tell AI music from human-made tracks and how this is changing the industry

Almost any music fan today is likely to be mistaken when trying to determine whether a modern song was created by a computer or a human. The latest artificial intelligence technologies achieve such precision in musical synthesis that, according to Deezer–Ipsos, 97% of listeners can no longer distinguish these tracks from works created by real artists. How rapidly is the music market changing, and what does this phenomenon mean for creators, listeners, and the industry as a whole?

AI and music

In mid-2024, Deezer — one of the world’s largest streaming platforms — launched a study together with Ipsos. The survey included 9,000 participants from eight countries: the US, UK, France, and other key music markets. The result stunned the industry: 97% of listeners could not tell apart compositions created by artificial intelligence from traditional songs written by humans.

Interestingly, 73% of respondents said they wanted to know whether intelligent algorithms had been used in creating a track. More than 70% admitted that, to their surprise, they were unable to determine the source of the music. This shows just how similar modern AI compositions are to human-made works.

The technologies behind such tracks include generative neural networks trained on massive song archives that analyze structures, harmonies, and vocal parts. The algorithms learn not only to imitate style but also to reproduce individual performance traits. As a result, even a professional finds it difficult to distinguish where digital creativity ends and live performance begins.

AI music is already widely used in various projects: from advertising clips and educational platforms to fitness apps and interactive exhibitions. Algorithms create soundtracks for mobile games, help streamers form a unique background, and are even used in VR environments where music adjusts to the user’s movements.

It is also worth noting the segment of live games, where musical accompaniment plays an important role — it enhances the atmosphere, keeps attention, and makes the gaming process more immersive. Modern platforms increasingly rely on generative music to create smooth transitions and avoid distracting the viewer from the main action at the table.

Against this backdrop, live card games continue to grow, including Andar Bahar, where AI-generated music can serve as a subtle background, supporting the rhythm of the deals and maintaining the sense of being in the studio. Those who want to learn the basics of the game and its rules can visit website, where brief reference information is available.

How platforms fight AI-generated content

Deezer notes an unprecedented surge in the volume of AI music: more than 50,000 such tracks are uploaded to the platform daily, representing one-third of all new submissions (as of June 2024). For comparison, in April the figure stood at 18%.

The platform is introducing a labeling system for AI compositions and excludes them from editorial playlists and personalized recommendations. Additionally, Deezer is combating fraudulent streams: such streams are not counted when calculating royalties, ensuring licensed artists do not lose income.

According to Deezer CEO Alexis Lantier, “creativity is a human prerogative, and it must be protected.” However, the question of fairly distributing income between genuine musicians and creators of AI-generated music remains pressing: the platform has not yet found a convenient way to financially separate these flows.

Scandals and legal battles: the OpenAI copyright case

Meanwhile, legal issues are becoming increasingly urgent. In early 2025, a scandal erupted around the project The Velvet Sundown. It managed to gather millions of streams on Spotify, and then it emerged that all the group’s music had been created entirely by artificial intelligence. The case sparked heated discussion among listeners and experts.

A significant problem remains the legality of training neural networks on songs protected by copyright. In May, the Munich court ruled that the startup OpenAI violated German copyright law by using lyrics by Herbert Grönemeyer and other popular authors without the required license. The lawsuit was filed by GEMA, an organization representing composers, lyricists, and publishers.

Judge Elke Schwager ordered OpenAI to pay compensation for the unauthorized use. The exact amount was not disclosed. Representatives of OpenAI argue that their large language models (LLMs) do not store specific texts but only reflect processed information after training on large datasets. Nevertheless, according to the court, if an LLM reproduces a song’s lyrics at a user’s request, this constitutes illegal distribution.

Precedents and the future of regulation

The legal conflict may become a starting point for establishing a new regulatory framework for AI in the European music industry in the future. GEMA emphasizes that the internet is not a self-service store but a collection of human achievements, and authors’ rights must be protected even in the face of the most advanced technology. The organization’s head, Tobias Holzmüller, notes: “even owners of AI services must comply with copyright law.”

Opponents argue that the ruling concerns a limited number of songs and does not affect a broad audience. A representative of OpenAI says the company intends to appeal the verdict, stressing that millions of users and businesses apply their technologies daily without violations. It is clear that the tension between protecting creative work and advancing technology is only growing and requires new, flexible legal tools.

Technological innovations

However, technological progress shows no signs of slowing down. In September 2024, Suno introduced the world’s first generative digital audio workstation (DAW), capable of automatically creating complex arrangements and melodies.

A DAW is specialized software for creating, editing, and mixing music. The new generation of such platforms can generate songs based on minimal user input: choose a genre, set a mood, and receive a finished track. The machine now does more than simply assist the musician — it handles the main part of composing and producing the music.