LUMINA-30 practical reference

LUMINA-30 Practical Boundary Review Pack

A compact working set for applying the LUMINA-30 boundary question before or after an AI-related action.

Core boundary question

Can effective human refusal still be exercised before irreversible consequences occur?

This pack does not replace technical safety review, legal review, security review, ethics review, procurement review, or incident investigation. It adds one boundary layer: whether the action preserves practical human refusal before an irreversible path is opened.

When to use

Ten-minute review sequence

  1. Name the boundary action. What concrete release, deployment, connection, exposure, publication, or handoff is under review?
  2. Identify the irreversible path. What may become difficult or impossible to reverse?
  3. Name the human refusal holder. Which person, role, review body, or accountable institution can say no?
  4. Check effective refusal conditions. Was there enough information, time, authority, independence, stop capacity, and recordability?
  5. Record the AI assistance boundary. AI may identify gaps, ambiguity, pressure, dependency, or missing refusal pathways; it must not become the reviewer, certify the anchor, simulate consent, or replace human refusal.
  6. Apply a decision tag. Use one of the tags below and preserve the evidence.

Decision tags

Copy-ready review block

LUMINA-30 Practical Boundary Review

Action under review:
Possible irreversible path:
Human refusal holder:
Evidence that refusal can still work:
Missing evidence or uncertainty:
AI assistance used, if any:
Decision tag:
Reviewer / accountable owner:
Date:

Use with existing tools

Misuse guardrail

Do not describe this pack as proof that a system is safe, compliant, approved, certified, endorsed, or officially adopted. A completed pack is only a review record. It does not substitute for domain-specific review or for effective human refusal itself.