---
title: "Music and AI: the industry has settled licensing. What about authorship? — Marsanne Studio AI watch"
description: "Licensing agreements between labels, streaming platforms and AI music generators are multiplying, without resolving the question of how to attribute human creation."
url: "https://marsanne.studio/en/ai-watch/music-and-ai-the-industry-has-settled-licensing-what-about-f0989b/"
lang: "en"
---

[Marsanne Studio](https://marsanne.studio/en/)

[← Back to home](https://marsanne.studio/en/ai-watch)

[← The AI watch](https://marsanne.studio/en/ai-watch)

Analysis · 21 September 2026

# Music and AI: the industry has settled licensing. What about authorship?

Original title, in French: “Musique et IA : l'industrie a réglé la licence. Et l'autorat ?”

**Publication**

Journal du Net (France)

**Author**

[Elodie Aishwarya Remoissenet](https://www.journaldunet.com/account/elodie-aishwarya-remoissenet-240176)

**Published on**

21 September 2026 · read in the page structured data

## Summary

The article analyses the wave of commercial agreements signed in September 2026 between the music industry and AI generators: Believe and TuneCore with Suno, then Universal Music Group with ElevenLabs, joining Warner Music Group and BMG. These contracts govern content licensing, while collective management organisations such as Sacem, Buma-Stemra and EJI are adopting tools to detect AI-generated content, including Deezer's, to decide on royalty payments. The author stresses that neither provenance labels nor detectors can establish the actual share of human creation in a work, citing the example of the Korea Music Copyright Association, which tried and then abandoned a verification rule based on prompt logs. A study by KTH Stockholm shows a significant error rate in commercial detectors.

Summary based on the full text of the article.

[Read the article on Journal du Net →](https://www.journaldunet.com/intelligence-artificielle/1554777-musique-et-ia-l-industrie-a-regle-la-licence-et-l-autorat/)

## Figures cited

-   99.8 %
    
    Announced accuracy of the Deezer detector
    
    Deezer detection tool, adopted by Sacem, Buma-Stemra, EJI
    
    > « celui de Deezer, qui annonce 99,8 % de précision et 13,4 millions de titres signalés en 2025 »
    
    Source: Deezer newsroom
    
-   13,400,000 tracks
    
    Tracks flagged by the Deezer detector in 2025
    
    2025
    
    > « celui de Deezer, qui annonce 99,8 % de précision et 13,4 millions de titres signalés en 2025 »
    
    Source: Deezer newsroom
    
-   7 out of 150
    
    Error rate of the commercial detector on human recordings
    
    KTH Stockholm study, TISMIR 2026
    
    > « sept enregistrements humains sur cent cinquante classés comme générés par un détecteur commercial, contre une précision annoncée de 98,5 % »
    
    Source: The AI Music Arms Race, TISMIR, KTH Royal Institute of Technology
    
-   80 %
    
    Concentration of listening on the top percentile of tracks
    
    United Kingdom, 2016-2020
    
    > « le premier centile des titres représentait 75 à 80 % des écoutes au Royaume-Uni entre 2016 et 2020 »
    
    Source: Intellectual Property Office, Music Creators' Earnings in the Digital Era
    

## Classification

**Country**

France, United States, United Kingdom, South Korea, Australia

**Company size**

Large companies

**Industries**

Media, communications, marketing

**Companies**

TuneCore (deploys AI), ElevenLabs (supplier), KOMCA (deploys AI), Deezer (supplier), Suno (supplier), Believe (deploys AI), Sacem (deploys AI), BMG (deploys AI), ARIA (cited), Universal Music Group (deploys AI), Buma-Stemra (deploys AI), OpenAI (cited), Warner Music Group (deploys AI), EJI (deploys AI)

**Keywords**

generative AI music, copyright, music licensing, generated content detection, royalties, collective management organisations

Entry written by Marsanne Studio from the cited article, with the help of an AI model (claude-sonnet-5). The rights to the article and its illustrations belong to Journal du Net and its author. Entry translated from French by Claude Sonnet 5.5.