Interoperability as Governance Infrastructure
The Principle
Effective regulation of complex, multi-actor systems depends on interoperable information exchange. The semantic and syntactic standards for how compliance information is represented and shared are governance infrastructure, as important as the rules themselves.
Why This Matters
The EU AI Act ecosystem involves multiple entities developing, providing, and deploying AI systems. Learning from implementation requires information to flow: between operators and authorities, between authorities across sectors and borders, between enforcement experiences and standards bodies, between all actors and affected stakeholders.
Without interoperable information models, each actor creates proprietary documentation. Learnings stay siloed. Pattern recognition across the ecosystem becomes impossible. Meta-learning, using observed outcomes to improve the rules, can’t function at scale.
How to Apply
-
Common foundational ontology: Develop shared terminology aligned with the regulation itself, provisional during initial interpretation, progressively extended as certainty increases
-
Semantic web standards: Use W3C standards (RDF, OWL, DCAT) that enable “alignment” and “translation” between different information models
-
Open data principles: Maximize transparency through open catalogues of compliance information (with appropriate confidentiality protections)
-
Data space infrastructure: Build shared infrastructure for cataloguing and accessing compliance information, similar to EU Open Data Portal approach
Where This Connects
This principle bridges:
- Knowledge Engineering: Ontology design, semantic modeling
- Information Architecture: Taxonomy development, findability
- AI Mechanisms: Technical standards for AI compliance
- Data Engineering: Data governance, interoperability protocols
The EU’s eProcurement Ontology offers a precedent: standardized concepts for modeling procurement data across the value chain, enabling common tooling and cross-border comparison.
Limitations
Interoperability standards take time to develop and adopt. Early-stage regulatory implementation may need to proceed with heterogeneous documentation formats, accepting that coordination will be imperfect. The goal is progressive improvement, not immediate perfection.
Related: 05-molecule—regulatory-learning-space-framework, 05-molecule—regulation-as-learning-framework, 05-atom—learning-activities-taxonomy