Ontological Knowledge Blocks: Executable Compliance and Profile-Based Validation for Trustworthy AI Systems
Researchers have developed Ontological Knowledge Blocks (OKBs), a new system designed to make AI compliance more automated and scalable. OKBs translate regulatory obligations into machine-checkable constraints using structured evidence graphs and SHACL validation rules. This programmable infrastructure allows for dynamic reconfiguration of governance profiles without altering the AI service code itself. Prototypes were tested in an HPC resource allocation scenario, demonstrating efficient validation times and the ability to manage different governance profiles. AI
IMPACT Automates AI compliance, potentially reducing manual audit burdens and enabling more scalable governance for critical AI systems.