Connection Note: OECD / AISI / EU AI Act Context

LUMINA-30 incident review public reference material.

Purpose

This note explains how LUMINA-30 may be positioned as a supplementary boundary reference within existing AI incident, safety, audit, and governance discussions.

Position

LUMINA-30 does not replace existing institutions, regulations, standards, or incident reporting systems.

LUMINA-30

It provides one missing boundary question:

LUMINA-30

Was meaningful human refusal effective before irreversible AI impact?

Incident Reporting Context

Existing incident reporting records what happened, what failed, and what harms or hazards occurred.

LUMINA-30 adds a boundary check:

LUMINA-30

Incident Response Context

Incident response often focuses on observation, containment, control, and recovery.

LUMINA-30 adds a prior boundary question:

LUMINA-30

Before recovery became necessary, was human refusal still effective?

Regulatory and Audit Context

Legal and regulatory frameworks may require human oversight, accountability, documentation, and risk management.

LUMINA-30 does not define those legal requirements.

LUMINA-30

It supports a supplementary post-incident question:

Was human oversight effective enough to preserve refusal before irreversibility?

Use this note as:


Current External Context (2026)

This section provides a current positioning layer for readers comparing LUMINA-30 with incident-monitoring, safety-report, and regulatory contexts. It does not claim that LUMINA-30 is endorsed, adopted, recognised, or required by any external institution.

OECD AI incidents and hazards context

The OECD AI Incidents and Hazards Monitor documents AI incidents and hazards as an evidence base for policymakers, practitioners, and other stakeholders. LUMINA-30 should not be described as an OECD tool, OECD standard, or OECD-aligned requirement.

OECD AI Incidents and Hazards Monitor

LUMINA-30 may be used only as a supplementary boundary question after or alongside incident documentation:

LUMINA-30

Did the review show that humans could still meaningfully refuse before irreversible impact?

EU AI Act / GPAI context

The EU AI Act sets obligations for providers of general-purpose AI models, including documentation, copyright-policy, and training-content-summary obligations for GPAI providers, with additional duties for GPAI models with systemic risk such as notification, risk assessment and mitigation, incident reporting, and cybersecurity protections.

EU AI Act

LUMINA-30 does not provide AI Act compliance, legal advice, certification, or regulatory approval. It may only be described as a non-binding boundary-reference question that can supplement governance or post-incident analysis.

LUMINA-30

The International AI Safety Report 2026 reviews scientific evidence on general-purpose AI capabilities and risks. LUMINA-30 does not replace AI safety reports, model evaluations, capability assessments, or technical safety research.

International AI Safety Report 2026

Its role is narrower:

It asks whether effective human refusal remained possible before the point at which AI-related impact became irreversible.
AI

Review Gap Mapping

External contextTypical focusLUMINA-30 supplemental questionNon-claim
OECD AI incident / hazard monitoringWhat happened, what harm or hazard occurred, and what patterns can be learnedWas human refusal still effective before irreversible impact?Not an OECD standard or OECD tool
EU AI Act / GPAI obligationsDocumentation, transparency, risk management, incident reporting, cybersecurityDoes the record show meaningful refusal authority before irreversibility?Not compliance, certification, or legal advice
International AI Safety Report / AISI-related safety contextCapabilities, risks, evidence gaps, safety researchDid safety review preserve a human refusal pathway before irreversible impact?Not a safety evaluation replacement or institutional endorsement
Research and governance discussionAnalysis, comparison, policy discussion, audit promptsWhat becomes incomplete if refusal-before-irreversibility is not checked?Not policy mandate, regulation, or approval regime
LUMINA-30
OECD AIOECD
EU AI Act / GPAI
International AI Safety Report / AISI