From a Registry to a Registry Ecosystem

For most of its history, European copyright law has been defined by the absence of formal registration requirements. But in a world where automated systems make split-second, copyright-relevant decisions about billions of works—filtering uploads, assembling AI training corpora, resolving reuse permissions—the lack of reliable, machine-readable rights information has become a critical infrastructure problem.

Today, we are excited to announce the publication of our final CommonsDB strategy paper: ‘From a Registry to a Registry Ecosystem’, written by Paul Keller (Open Future) and Sebastian Posth (Liccium). The paper tackles the most pressing use case in the current debate: the text and data mining (TDM) opt-out under Article 4(3) of the CDSM Directive, and the compliance challenges it creates for the AI industry under the AI Act.

What CommonsDB Has Demonstrated

Over the past 18 months, the CommonsDB pilot has successfully built a working public registry of rights information for public domain and openly licensed works. By testing our architecture without the friction of contested rights claims, we've demonstrated that purpose-built, ISCC-based registries work technically, operationally, and at scale.

Key takeaways from the pilot include:

• Scale and Success: The registry now holds over 6.5 million verified declarations, surpassing our initial target.
• Content-Based Identification: ISCC codes derived directly from media files allow anyone to independently generate an identifier without relying on fragile embedded metadata that often gets stripped in circulation.
• Federation in Action: Using an open-source peer-to-peer distributed layer, registry entries are published and synchronized directly between nodes, allowing high-volume users (like AI developers) to hold and query local copies.
• Verifiable Trust: Declarations are bound to works via ISCCs, signed by declarers, and securely anchored in eIDAS-compliant verifiable credentials.

The Future is Federated

Recent studies commissioned by the EUIPO and DG Connect agree that an EU-level registry of TDM opt-outs is both necessary and technically feasible. However, these studies often lean toward centralized architectures where all records flow into a single canonical index.

As our paper argues, the question is no longer whether a registry can be built, but what happens when many registries exist. The emerging landscape is not a single, monolithic database, but an ecosystem of registries maintained by different entities.

A centralized aggregation model introduces single points of failure, gatekeeping risks, and bottlenecks for high-volume compliance checking. Based on the experiences with  CommonsDB our paper proposes a federated ecosystem of registries In this model, reservation records remain in purpose-built, independently operated registries governed by the communities that populate them. A signed, machine-readable list maintained by a public certifier defines which registries are compliant, and records are synchronized over a common peer-to-peer protocol.

The Role of Institutions

All of this points to a more decentralized bottom up future for the emerging European copyright infrastructure. To make such an ecosystem a reality, the EUIPO should take a coordinating role. Rather than conceptualizing a new central database that absorbs existing registries, the EUIPO should focus on its unique position in the ecosystem: it should act as standards-setter and registry-certifier, operate the discovery and aggregation layers that connect the ecosystem, and migrate its existing databases onto this shared architecture.

To succeed, registries are better built and governed as close as possible to the communities they serve.