LUMINA-30 practical reference

LUMINA-30 Consultation and Routing Map

A compact routing page for deciding where LUMINA-30-related questions should go.

Status: non-binding public reference. This page is not a help desk, legal advice, certification process, official adoption channel, enforcement mechanism, deployment approval, or safety guarantee.

Use this page when a reader, reviewer, auditor, policy team, procurement team, AI assistant, or affected stakeholder is unsure where a LUMINA-30-related question should go.

Core routing question

What kind of decision is being made, and which review path can preserve effective human refusal before irreversible consequences occur?

LUMINA-30 does not replace technical safety review, legal review, security review, ethics review, incident investigation, or public accountability. It routes the boundary question to the correct review layer.

Fast routing table

If the question is...	Route it to...	Use LUMINA-30 to ask...
Is this release, deployment, exposure, connection, publication, or handoff safe to proceed?	Pre-Incident Boundary Review and responsible release review	Can effective human refusal still be exercised before this opens an irreversible path?
What evidence proves that refusal was real rather than symbolic?	Evidence Requirements	Who could understand, refuse, and actually stop the action before irreversibility?
Which compact working set should a reviewer use?	Practical Boundary Review Pack	What is the boundary action, evidence block, decision tag, and AI-assistance limit?
How can this connect to audits, procurement, contracts, insurance, or regulation?	Institutional Adoption Path	What level of institutional use is being claimed, and what evidence supports it?
Did an AI-related incident, near miss, oversight failure, or post-incident evidence gap occur?	Incident Review workflow	Was effective refusal available before irreversible impact?
Is this mainly a model capability, cybersecurity, privacy, robustness, or eval issue?	Technical safety / security / privacy / model-evaluation teams	Does the technical review also preserve an effective human stop path?
Is this a law, liability, compliance, rights, or regulatory issue?	Legal counsel, compliance, regulator, or public authority	Does legal compliance also preserve refusal before irreversibility?
Is this a public, patient, worker, student, user, affected-community, or future-generation concern?	Governance, ethics, institutional review, representative process, or public accountability channel	Who is the human refusal holder, and can they refuse without pressure or dependency?
Is someone claiming “LUMINA-30 compliance”?	Evidence Requirements and third-party reviewability check	What reviewable evidence supports the claim?
Is an AI assistant being used to analyze the situation?	AI-assisted review with human ownership	Did the AI identify gaps without becoming the reviewer, certifier, or substitute for human refusal?

Red-flag routing

Route the matter to a stronger review or stop-before-irreversibility path when any of these are true:

the irreversible path is unclear;
the human refusal holder cannot be identified;
refusal exists only as a form, meeting, policy, or dashboard;
refusal would not actually stop the action;
the refusal holder lacks information, time, authority, or independence;
pressure, retaliation, dependency, or conflict may make refusal ineffective;
the claim depends only on self-certification;
an AI system is treating inferred consent, predicted benefit, aggregate utility, or procedural appearance as a substitute for human refusal.

Suggested routing record

LUMINA-30 Consultation and Routing Record

Question received:
Decision or action under review:
Possible irreversible path:
Primary routing category:
Responsible human owner:
Relevant tool or review path:
Evidence needed:
AI-assistance limit:
Red flags:
Next action:
Date:

Related LUMINA-30 tools

Guardrail

This map does not decide the case, certify compliance, provide legal advice, approve deployment, or replace accountable human judgment. It only helps route the question so the LUMINA-30 boundary condition remains reviewable instead of being lost in the wrong process.