According to an article by Jean Albert, president of the Agency for the Protection of Programs (APP).
Key takeaways on music and AI
- Music production using AI raises issues related to visibility and compensation.
- Music streaming platforms and music awards have different eligibility requirements.
- The use of works by AI models may give rise to infringement lawsuits.
- To protect against this and defend one’s copyright, keeping a running record of the creative process helps document not only ownership but also the individual’s contribution to the work.
- A timestamp provides proof of the date and integrity of the data.
Safeguarding Musicians' Income and Visibility
In July 2026, Deezer reported that it was receiving nearly 90,000 tracks generated entirely by AI every day.
A surge in AI-generated music that limits the visibility of human artists
During peaks recorded in June, this content accounted for more than 50% of new daily music releases. However, its share of total streams remains between 1% and 3%. This low share is partly attributable to the fact that Deezer, when it is able to identify these tracks, limits their visibility on the platform by excluding them from its algorithmic recommendations and editorial playlists. [1]
Beatport is taking it a step further by taking action as soon as content is uploaded to its platform. On August 12, 2026, the platform announced the expansion of its partnership with Beatdapp to detect and block tracks that are entirely or predominantly generated by AI. AI assistance is still permitted when the result remains predominantly human. [3]
The rise of AI-generated music tracks therefore poses a risk to artists’ visibility on music streaming platforms. But another trend poses an equally direct threat to the platforms themselves: streaming scams.
A wiretapping scam involving AI agents
Since artists are paid based on the number of times their songs are streamed, some have created AI agents that listen to the music and thereby generate enough streaming volume to pay their creators. Deezer reports that it has been able to identify up to 85% of these fake streams and exclude them from royalty calculations. [2]
Determine which contributions can be recognized and compensated
Faced with this surge of AI in music, players in the music industry have begun to implement recommendations and measures—some more radical than others—regarding the acceptance of AI-generated music.
AI is bringing about a change in the criteria used to rank and reward musical achievements
In Australia, Josh Fawaz’s cover of “Like a Prayer” was a huge hit. After reaching number one on Australian radio charts, the song climbed to number four on the ARIA (Australian Recording Industry Association) charts. This success has raised questions, particularly regarding the extent to which AI was used in the adaptation of this song. The controversy prompted the artist to clearly state in the credits that the vocals and drums were generated by AI.
This case raises the question of what is being recognized in a ranking or award: commercial success, or the person behind the creation? [4]
This incident prompted ARIA to announce its new rules for classifying musical recordings on August 25, 2026. Under these rules, a lead vocal or a major instrumental part generated by AI renders the recording ineligible. Certain secondary contributions and technical aids remain permitted. In cases of credible doubt, the organization may request evidence to substantiate the track’s eligibility. [6] [7]
Evolving Principles
The same questions arise regarding the attribution of intellectual property rights to works created using AI. We are also seeing that rulings are evolving.
For example, in March 2025, South Korea’s KOMCA, one of the country’s leading copyright collective management organizations, had completely excluded works involving AI from its registration process. On August 3, 2026, it resumed accepting such works for registration on the condition that a substantial human contribution to the writing, structure, or arrangement could be demonstrated . [ 8] [9]
Protecting works used or adapted by AI models
The authors’ defense also focuses on pre-existing works used by AI models for training. Here, too, the debate is raging, and it is often difficult for an author to prove that the AI “copies” their works. The GEMA v. Suno lawsuit provides a concrete illustration of the issues at stake.
On July 31, 2026, in a case pitting GEMA (a German collective management organization for the copyrights of composers, lyricists, and music publishers) against Suno (an AI-assisted music creation platform), the Munich Regional Court ruled in favor of GEMA. It found that Suno was guilty of copyright infringement regarding six original and protected musical works. It found infringements related to their reproduction during training, their storage in the models, and their reproduction in the generated results; this ruling was based on an analysis of the prompts, which clearly demonstrated the intent to create songs “in the style of.” [10]
This case highlights the importance of being able to identify which aspects of a work’s creation are the result of human creativity and which are the result of AI, and of being able to prove it by archiving all the elements that led to the final work. This practice will make it possible to date the work’s creation, prove the human contribution to its creation, and defend one’s rights in the event of a dispute.
What evidence can authors provide to protect their works?
In all jurisdictions that are signatories to the Berne Convention, copyright arises by virtue of creation alone, without any requirement for registration. Protection requires that a work be expressed in an identifiable and original form, reflecting the author’s free and creative choices. A mere idea is not protectable. [11] [12]
Who can be the owner of a copyright?
The author must be a natural person or a legal entity. A machine, on the other hand, cannot be considered an author. In the event of a dispute over these rights or an infringement, one must be able to provide proof of ownership, the date of creation, and originality. [13] [14]
The Role of Timestamping in Copyright Protection
Dating helps establish a definite date of creation for the work and a chronology of its evolution, provided that this evolution is traceable and documented. The method of proving the date is flexible, but it is best to ensure that the proof of the date cannot be easily disputed. This is whata qualified timestamp provides.
According to Article 41 of the eIDAS Regulation,a qualified electronic time stamp is presumed to accurately record the date and time, as well as to ensure the integrity of the associated data. It provides certain and indisputable proof of the existence of a version at a given point in time. However, it does not, on its own, certify the originality or authorship of the document. [15]
Nor does a timestamped creation, on its own, suffice to document the extent of human contribution to the creative process: compositional choices, modifications, and elements that were generated but later discarded. As our article on copyright and proof of human creation explains, the issue centers on the creative contributions that were actually made.
Linking the different versions of a work to the history of its creation helps to better substantiate these contributions. The second article in this series will show how a musical creation journal can document the human effort involved and organize the evidence throughout the project.
Sources
[1] Deezer, data on AI-generated music streams and plays, July 21, 2026
[2] Deezer, Detection and Demonetization of Fraudulent Streams, January 29, 2026
[3] Beatport, Partnership with Beatdapp Extended, August 12, 2026
[4] ABC News, AI credits for Josh Fawaz’s *Like a Prayer*, July 17, 2026
[5] IFPI, Chart Eligibility Guidelines, July 30, 2026
[6] ARIA, announcement of new rules, August 25, 2026
[7] ARIA, Frequently Asked Questions About AI Regulations
[8] Music Business Worldwide, Changes to KOMCA Rules, August 5, 2026
[9] Yonhap, KOMCA Withdraws New Rules, August 25, 2026
[10] Munich Regional Court I, GEMA v. Suno, Press Release No. 16 of July 31, 2026, Case No. 42 O 763/25
[11] Intellectual Property Code, Article L111-1
[12] CJEU, Cofemel, September 12, 2019, C-683/17, paragraphs 29–34
[13] Court of Cassation, 1st Civil Chamber, January 15, 2015, Appeal No. 13-23.566
[14] Intellectual Property Code, Article L113-5
[15] eIDAS Regulation, Article 41, consolidated version of October 18, 2024
Disclaimer of liability
The opinions, presentations, figures and estimates presented on the website, including the blog, are intended for information purposes only and should not be construed as legal advice. For legal advice you should contact a legal professional in your jurisdiction.
The use of the content of this website, including the blog, for commercial purposes, including resale, is prohibited without prior authorization from Evidency. The request for permission must specify the purpose and extent of reproduction. For non-commercial purposes, all material in this publication may be freely quoted or reprinted, but acknowledgement is required, together with a link to this website.
