Public reference ontology • stewarded by AISYSTEMS s. r. o.

AI Core

A public, descriptive reference ontology for AI audit interoperability. Designed for version stability, integrity anchoring, and long-term reproducibility of technical evidence.

Scope

AI Core defines descriptive terminology for AI audit and observation contexts. It provides a stable conceptual vocabulary used to anchor audit artifacts and evidence.

Non-goals: AI Core does not prescribe implementation behavior, does not define audit methodology, and has no regulatory authority.

How it is used

Audit systems reference a specific AI Core release in their AuditPlan and evidence outputs using a version + cryptographic hash (Ontology Anchor). This prevents semantic drift and enables reproducibility years later.

Canonical release artifacts are published under /vX.Y/, each containing a human-readable document (index.html) and the machine-readable ontology (ontology.jsonld).

Principles

Descriptive: Non-normative definitions of terms.
Versioned: Every release is immutable and published separately.
Integrity-anchored: Releases are referenced via sha256.
Public: Free for use by any third-party audit or regulatory tool.

Stewardship

AI Core is developed and stewarded by AISYSTEMS s. r. o. It may be referenced by audit systems, including AISYSTEMS SENTINEL, while remaining independent and descriptive in nature.