LUMINA-30 is a boundary framework for determining whether human refusal remained effective before irreversible AI autonomy emerged.
Why this matters here
LUMINA-30 does not replace capability evaluation, alignment research, or safety evaluation.
It adds a separate boundary question:
Did meaningful human refusal remain effective before irreversibility?
Position relative to existing evaluation
Capability: what the system can do
Alignment: how the system behaves relative to intended aims
Safety evaluation: where failure may occur
LUMINA-30: whether humans can still refuse effectively before irreversible impact
Research value
works across paradigms
useful for post-hoc validation
connects technical evaluation to institutional review
treats uncheckable refusal conditions as invalid rather than merely uncertain
Conclusion
LUMINA-30 does not compete with alignment or capability evaluation. It asks a different boundary question: whether human refusal remains effective before irreversibility.