Special issue Journal of Intellectual Property Law & Practice about music metadata

This paper shows that music industry and EU initiatives at the start of the online era for music consumption between the early 2000s and the early 2010s, aiming for centralized copyright databases, failed in part due to misaligned remuneration systems and economic priorities.

Some challenges present since those early years have remained, while new ones have emerged with the advent of music streaming, and more recently also with generative AI (GenAI) music tools and services. Decentralized solutions also still have to grapple with metadata design challenges for attaining music metadata interoperability, generally with regard to domain specificity, granularity and provenance.

The transparency obligations in Articles 50 and 53(1)(d) of the AI Act do not provide sufficient practical, enforceable rules that can improve metadata interoperability or copyright attribution for GenAI music in the (European) music industry. The explanations and guidance given in the First Draft Code of Practice in relation to Article 50 or the Explanatory Notice and Template for Article 53 do not sufficiently fill those gaps either.

This article examines the persistent crisis of music metadata and its implications for discoverability, attribution and remuneration. It distinguishes four key categories of metadata and argues that rights-related metadata is both the most economically consequential and the hardest to produce and maintain, because it encodes complex, jurisdiction-specific and dynamically changing relationships among creators, intermediaries and users.

The paper shows how legal theory and tradition, the fragmentation of copyright and licensing and the proliferation of heterogeneous intermediaries have jointly prevented the emergence of comprehensive public or private registries. It reviews three major families of attempted solutions: centralized institutional initiatives (eg CMO-led registries), open peer-produced datasets (eg MusicBrainz, Discogs), and blockchain/smart contract–based projects, and explains why each has, so far, failed to deliver an authoritative, rights-inclusive global repertoire database. The analysis situates these failures within broader industry transformations: the democratization of music production, the financialization of rights, the consolidation of power at a few digital service providers and the dual role of AI as both a tool for metadata generation and a driver of repertoire inflation through generative systems.

The paper also discusses the 2025 leak of Spotify’s metadata and audio catalogue via Anna’s Archive, interpreting it as a paradigmatic turning point. The sudden availability of the largest publicly accessible, rights-linked music metadata set in history raises legal, institutional and sustainability questions. The article concludes with recommendations towards a more equitable and efficient music data commons—not shying away from the use of the leaked Spotify dataset.

Music metadata—credits, identifiers, language labels, territorial tags and genre descriptors—functions as the operative infrastructure of streaming. It shapes what becomes searchable, recommendable, charted and remunerated. This article argues that metadata is therefore not a neutral technical resource but a constitutional site where structural inequality is produced or mitigated. When metadata is sparse, standardized around dominant markets, or mis-specified, the resulting visibility and remuneration deficits disproportionately affect minority-language repertoires, music from smaller territories, field recordings and traditional archives, and women and non-binary creators.

The article situates these ‘structural metadata harms’ within the EU’s fundamental-rights framework, contending that Article 22 CFR (respect for cultural and linguistic diversity), read together with Articles 11 (freedom of expression), 13 (artistic freedom), 17(2) (IP), 21 (non-discrimination) and 23 (gender equality), constrains and guides metadata governance. Drawing on CJEU rights-balancing and ECtHR doctrines of positive obligations and indirect structural discrimination, it develops the claim that EU regulatory and standard-setting choices must secure the practical and effective enjoyment of cultural visibility and equal rights-realization online.

Music recommender systems: Copyright law secures the lawful availability of content on platforms, management of rights, or remuneration for use, but it remains indifferent to the allocation of cultural attention. Music recommender systems structure the conditions under which copyright can generate cultural and economic value. Their optimisation logics shaped by metadata quality, engagement metrics, and commercial incentives, influence which works circulate and which remain peripheral.

Doctrinal blind spot: This article argues that the resulting condition of being represented but not heard exposes a doctrinal blind spot in the copyright system and engages constitutional concerns rooted in artistic freedom, freedom of expression, and the EU’s commitments to cultural diversity. When music discovery becomes infrastructural rather than incidental, cultural participation can no longer be assessed in terms of consumer access to content or platform availability but must also account for artist audibility. The right to be heard: The article reconstructs the right to cultural participation as a relational and infrastructural right to be heard pointing to the need for norms that prevent systemic exclusion where private infrastructures function as unavoidable gateways to cultural encounter. Drawing inspiration from the prominence requirement in the AVMS Directive, it argues that a prominence-based regulatory approach offers a proportionate response to governing attention in markets where exposure is no longer a by-product of availability but a function of design. By focusing on objectives rather than outputs and by leaving technical implementation to platforms, such an approach preserves the essence of the freedom to conduct a business while acknowledging that entrepreneurial autonomy cannot extend to the systematic foreclosure of cultural audibility.

With the work notification mechanism in Article 17(4)(b) and the opt-out mechanism in Article 4(3) of the 2019 Directive on Copyright in the Digital Single Market, EU copyright legislation seeks to strengthen the position of composers, performers, record labels and music publishers vis-à-vis online platforms and AI developers.

From the perspective of copyright data governance, however, these provisions may have unintended side effects. Instead of strengthening the position of artists and music companies, they set in motion metadata streams from the music industry to providers of online platforms and AI developers. Ultimately, these metadata fortify the data hegemony of large technology companies. They may increase the dependence of artists and music companies on big tech platforms and AI systems for the distribution of content.

To counterbalance the data hegemony of platforms and AI providers, it is advisable to establish a comprehensive EU music metadata infrastructure: a central, open data repository that bundles Article 17(4)(b) work notifications and Article 4(3) opt-out statements which are enriched with descriptive and rights clearance metadata. Once these work notifications and opt-out declarations are in parallel collected and pooled in a central EU music metadata repository, the resulting accumulation of music data can lead to a promising data reservoir capable of competing with the data collections held by online platforms and AI trainers.

Music metadata is central to remuneration accuracy and cultural visibility in AI- and platform-mediated markets, yet Europe’s landscape remains fragmented due to territorial rights management, divergent identifiers, and proprietary silos. The article reframes the resulting issues — unmatched royalties, inefficient licensing and distorted discoverability — as problems of governance and institutional design rather than purely technical ones.

It evaluates whether a decentralized data space architecture, aligned with the EU’s Common European Data Spaces initiative, offers a viable alternative to repository-based models. Drawing on EU data and digital legislation (notably the Open Data Directive, Data Governance Act, and Data Act) and using the European Health Data Space as a reference point, it identifies key legal mechanisms, including interoperability obligations, constraints on unfair contractual terms, and intermediary governance structures.

The analysis highlights persistent tensions between EU data law and regimes such as copyright and trade secrets, and assesses the extent to which the current framework can enable governed interoperability for music metadata, pointing to areas where more targeted, sector-specific intervention may be required.