By signing up, you agree to receive emails every time the Deezer Newsroom is updated. You can unsubscribe at any time by clicking the link at the bottom of each email.
Paris, September 11, 2025 – Deezer, the global music experiences platform, is receiving over 30,000 fully AI-generated tracks every day – accounting for more than 28% of the total daily delivery. Deezer’s AI detection tool has been in place since the beginning of the year, enabling the company to track a steady increase of fully synthetic content on the platform. In June, Deezer became the first (and so far only) music streaming platform to explicitly tag AI-generated music.
“Following a massive increase during the year, AI music now makes up a significant part of the daily track delivery to music streaming and we want to lead the way in minimizing any negative impact for artists and fans alike,” said Alexis Lanternier, CEO, Deezer. “Our approach is simple: we remove fully AI-generated content from algorithmic recommendations and we don’t include it in editorial playlists. This way we ensure the impact on the royalty pool remains minimal, while providing a transparent user experience. And most importantly, we continue to fight fraudulent activity, which is the main driver behind uploading fully AI generated content.”
In January, Deezer reported that roughly 10 % of all content delivered to the platform was AI generated. This number increased to 18% in April and has now reached a record breaking 28%.
*** ENDS ***
Notes to editors:
Deezer is committed to protecting the rights of artists – The increase in fully AI generated content in music streaming comes at a time of growing concerns about AI companies training their models with copyrighted material, and governments potentially diminishing copyright laws to facilitate AI development. Deezer is committed to protecting the rights of artists and creators, and remains the only streaming platform to sign the global statement on AI training.
Deezer’s AI music detection tool sets an industry standard, with the ability to detect 100 % AI-generated music from the most prolific generative models – such as Suno and Udio, with the possibility to add detection capabilities for practically any other similar tool as long as there’s access to relevant data examples. Not only that, Deezer has made significant progress in creating a system with increased generalizability, to detect AI generated content without a specific dataset to train on.
AI is a critical challenge for the music industry – According to a study conducted by CISAC and PMP Strategy, with participation from key industry players (including Deezer), nearly 25% of creators’ revenues are at risk by 2028, which could amount to as much as €4 billion by that time. This represents a colossal, even critical, challenge for the music creation sector as a whole. https://www.cisac.org/services/reports-and-research/cisacpmp-strategy-ai-study
AI and fraud – Although fully AI-generated music currently accounts for only a small fraction of streams on Deezer — approximately 0.5% — it’s evident that the primary purpose of uploading these tracks to streaming platforms is fraudulent. Deezer has found that up to 70% of the streams generated by fully AI-generated tracks are in fact fraudulent. When detecting stream manipulation of any kind, Deezer excludes the streams from the royalty payments.
AI tracks are removed from recommendations – All 100% AI generated songs are automatically removed from algorithmic recommendations and are not included in editorial playlists. This is a first step in making sure that these tracks don’t dilute the royalty pool in any significant way. Potential future actions, including updating our supplier policy and removing/demonetizing content need to be based on careful consideration.
Two new patents – In December 2024, Deezer applied for two patents for its AI Detection technology, focused on two different methods of detecting unique signatures that are used to distinguish synthetic content from authentic content.
Press Contact Deezer
Jesper Wendel
jwendel@deezer.com
On music streaming services, listening sessions are often composed of a balance of familiar and new tracks. Recently, sequential recommender systems have adopted cognitive-informed approaches, such as Adaptive Control of Thought—Rational (ACT-R), to successfully improve the prediction of the most relevant tracks for the next user session. However, one limitation of using a model inspired by human memory (or the past), is that it struggles to recommend new tracks that users have not previously listened to. To bridge this gap, here we propose a model that leverages audio information to predict in advance the ACT-R-like activation of new tracks and incorporates them into the recommendation scoring process. We demonstrate the empirical effectiveness of the proposed model using proprietary data, which we publicly release along with the model’s source code to foster future research in this field.
We explore a novel use case for Large Language Models (LLMs) in recommendation: generating natural language user taste profiles from listening histories. Unlike traditional opaque embeddings, these profiles are interpretable, editable, and give users greater transparency and control over their personalization. However, it is unclear whether users actually recognize themselves in these profiles, and whether some users or items are systematically better represented than others. Understanding this is crucial for trust, usability, and fairness in LLM-based recommender systems.
To study this, we generate profiles using three different LLMs and evaluate them along two dimensions: self-identification, through a user study with 64 participants, and recommendation performance in a downstream task. We analyze how both are affected by user attributes (e.g., age, taste diversity, mainstreamness) and item features (e.g., genre, country of origin). Our results show that profile quality varies across users and items, and that self-identification and recommendation performance are only weakly correlated. These findings highlight both the promise and the limitations of scrutable, LLM-based profiling in personalized systems.
Natural language interfaces offer a compelling approach for music recommendation, enabling users to express complex preferences conversationally. While Large Language Models (LLMs) show promise in this direction, their scalability in recommender systems is limited by high costs and latency. Retrieval-based approaches using smaller language models mitigate these issues but often rely on single-modal item representations, overlook long-term user preferences, and require full model retraining, posing challenges for real-world deployment. In this paper, we present JAM (Just Ask for Music), a lightweight and intuitive framework for natural language music recommendation. JAM models user–query–item interactions as vector translations in a shared latent space, inspired by knowledge graph embedding methods like TransE. To capture the complexity of music and user intent, JAM aggregates multimodal item features via cross-attention and sparse mixture-of-experts. We also introduce JAMSessions, a new dataset of over 100k user–query–item triples with anonymized user/item embeddings, uniquely combining conversational queries and user long-term preferences. Our results show that JAM provides accurate recommendations, produces intuitive representations suitable for practical use cases, and can be easily integrated with existing music recommendation stacks.
This paper has been accepted for publication in the proceedings of the 19th ACM Conference on Recommender Systems (RecSys 2025).
Paris, September 4, 2025 – Deezer (Paris Euronext: DEEZR), the global music experiences platform, has been named to Newsweek’s list of the world’s most trustworthy companies in 2025. The ranking includes 1000 companies in 20 countries, and is based on surveys of 65,000 people as well as extensive social media listening. Deezer is the only music streaming company on the list and is ranked in the Media & Entertainment category, along with music industry heavyweights such as Warner Music Group and Universal Music Group.
“I’m incredibly proud that Deezer is considered one of the most trustworthy companies in the world,” said Alexis Lanternier, CEO of Deezer. “It shows that our commitment to fairness and transparency in the music business resonates not only with our industry peers, but also with music fans across the world. Our investment in R&D for fairer artist payments, fraud prevention, and most recently AI tagging, has put us in the global spotlight and showcases that we genuinely care about the future of music. This, in combination with delivering world class music experiences, both in the app and through our fan events, has led us to this achievement.”
The full list and a rundown of the survey methodology can be found with the following link https://rankings.newsweek.com/worlds-most-trustworthy-companies-2025
*** ENDS ***
Press Contact Deezer
Jesper Wendel
jwendel@deezer.com