OpenAIRE presented OpenREL on 4 May 2026 at FOR2026 in Munich, as part of the session “Introducing OpenREL: Rights Expression Languages for Open Science and International Data Spaces - A Practitioners’ Approach”. The session was delivered by Prodromos Tsiavos and Melios Katsamakis, within the EOSC Beyond project.

The presentation opened by highlighting a fundamental limitation in how digital rights are currently expressed in research. Most datasets are published with licence labels such as CC BY or CC BY-NC, or simply marked as “open access”. While these are understandable by humans, they are not actionable by systems. In a federated environment such as EOSC, where access and reuse increasingly happen at scale, relying on human interpretation is no longer viable. In addition, traditional licences do not capture key dimensions such as data protection constraints, ethical conditions, embargo periods, or role-based access. This creates a gap between what researchers intend and what machines can actually understand and act upon.

This gap is becoming more critical as research consumption increasingly involves machines as well as humans. AI agents and automated workflows are becoming important consumers of research outputs, yet they cannot reliably interpret licence text or unstructured usage conditions. As a result, data may be reused without sufficiently structured governance, transparency, or accountability. The presentation positioned this as an infrastructure challenge that requires a systematic response.

OpenREL was introduced as a policy layer designed to address this need. Built on ODRL, a W3C standard, OpenREL translates licences, usage conditions, and governance rules into structured, machine-readable policies that can be evaluated automatically by systems and AI agents. The approach extends across multiple dimensions of rights, including intellectual property, data protection, ethical considerations, and contractual conditions, and produces aligned outputs that are human-readable, legally meaningful, and machine-actionable.

The system operates across two complementary flows. On the authoring side, data providers and stewards define policies through a structured interface, expressing permissions, prohibitions, and obligations without requiring knowledge of the underlying technical model. These policies are generated in a machine-readable format, assigned persistent identifiers, and linked to datasets. On the runtime side, users or automated systems submit structured access requests, specifying identity, action, and purpose. OpenREL evaluates these requests against the applicable policy and returns a deterministic outcome, PERMIT, CONDITIONAL, or DENY, always accompanied by a justification. Where structured policies are not available, the system can rely on existing licence information, and where no information is present, it escalates to human review. Decisions can also be logged to support transparency and accountability.

In terms of progress, the work has moved from an initial analytical phase into early implementation. Efforts have focused on defining a structured approach to rights representation, identifying suitable technical foundations, and translating these into practical components. Current activities include the development of use cases, the structuring of a Knowledge Base, and the design of policy models and templates that can support real-world scenarios.

The next phase will focus on advancing implementation, finalising core components such as the Knowledge Base, API, and user-facing tools, and validating the approach with relevant stakeholders, including data providers and research infrastructures. The objective is to ensure that the approach is practical, scalable, and aligned with the needs of the EOSC ecosystem.

The presentation concluded by situating OpenREL within a broader transition in Open Science. While Open Access has significantly improved the availability of research outputs, the next step is enabling their controlled, transparent, and trustworthy reuse by systems and AI at scale. This requires rights to be not only declared, but structured, machine-readable, and operational. OpenREL contributes to this transition by providing a policy-aware layer for data access and reuse within EOSC, helping research infrastructures move towards more accountable, interoperable, and AI-ready Open Science services.