Ontological Knowledge Blocks for Trustworthy AI Governance
A recent publication presents Ontological Knowledge Blocks (OKBs), a programmable governance framework aimed at automating compliance for AI systems within essential digital infrastructures. Existing compliance strategies depend on written documentation, fixed checklists, and manual assessments, which lack scalability. OKBs transform regulatory requirements into machine-verifiable constraints on structured evidence graphs. This formalization connects normative responsibilities to an RDF/OWL concept schema, executable SHACL validation rules, specific evidence demands, and PROV-O provenance links. A deterministic regulatory compiler converts structured Intermediate Representation (IR) records into modular KB components, facilitating governance reconfiguration based on profiles. The study is available on arXiv with the identifier 2605.23297.
Key facts
- OKBs compile regulatory obligations into machine-checkable constraints
- Current compliance is documentation-centric and manual
- OKBs use RDF/OWL concept schema, SHACL rules, and PROV-O provenance
- A deterministic regulatory compiler translates IR records into KB modules
- Paper available on arXiv:2605.23297
- Focus on transparency, accountability, fairness, traceability
- Designed for critical digital infrastructure
- Enables profile-based governance reconfiguration
Entities
Institutions
- arXiv