AI Core – Reference Ontology v1.1 (LOCKED)

0. Scope

This document provides a public, descriptive reference ontology for the term AI Core as used in technical, audit, and research contexts.

This document does not prescribe implementation behavior, assess correctness, or provide compliance, certification, or governance outcomes.

1. Foundational Observations

Non-Interference Observation: AI Core components operate without modifying the runtime state of the observed system.

Traceability Observation: Outputs are associated with identifiable version references and temporal metadata.

Reproducibility Observation: Identical inputs under identical core references yield identical technical outputs in deterministic designs.

2. Formal Definitions

Term	Canonical URI	Description
`AI Core`	https://aicore.sk/def/ai-core	A conceptual core component associated with observation, inspection, or evaluation of AI system behavior.
`Audit Core`	https://aicore.sk/def/audit-core	A conceptual core component commonly used to perform deterministic inspection of AI-related data or outputs.
`Evidence Layer`	https://aicore.sk/def/evidence-layer	A logical layer responsible for recording technical observations in immutable or append-only form.

3. Separation of Domains

Application Domain: Invocation context.
Audit Domain: Observation context.
Source Domain: Observed AI system.

4. Evidence Representation (Illustrative)

{
  "cid": "unique-evidence-id",
  "meta": {
    "protocol": "aicore.sk/v1.1",
    "core_reference": "sha256:...",
    "timestamp": "2026-02-22T16:00:00Z"
  },
  "payload": {
    "observation": "raw_data",
    "result": "deterministic_output"
  },
  "proof": "cryptographic_signature_or_hash"
}

5. Referencing This Ontology

Whole document: https://aicore.sk (Version 1.1, LOCKED, February 2026)

Machine-readable version: https://aicore.sk/v1.1/ontology.jsonld

6. Stewardship

This ontology is stewarded for consistency and long-term availability. Stewardship does not imply authority over interpretation or implementation.

7. Versioning

Version v1.1 extends v1.0 by adding descriptive terminology related to distributed audit systems. Previous versions remain accessible.

8. What AI Core Is Not

A certification scheme
A compliance assessment service
A regulatory authority
A ranking or scoring system
An enforcement mechanism
A normative standard

9. Distributed Audit Vocabulary (Descriptive Extension)

Term	Canonical URI	Description
`Audit Plan`	https://aicore.sk/def/audit-plan	A structured, version-bound specification describing deterministic observation steps and constraints used during an audit execution context.
`Runner`	https://aicore.sk/def/runner	A distributed execution component responsible for performing observation steps and producing evidence artifacts.
`Evidence Manifest`	https://aicore.sk/def/evidence-manifest	A canonical serialized index describing artifacts produced during audit execution.
`Receipt`	https://aicore.sk/def/receipt	A cryptographically signed acknowledgment that a manifest has been received within an audit control context.
`Escrow Release`	https://aicore.sk/def/escrow-release	A controlled disclosure event referencing previously indexed artifacts.
`Determinism Parameters`	https://aicore.sk/def/determinism-parameters	Structured parameters that constrain execution reproducibility such as seed values and step ordering.