Article 9. Can a System That Cannot Lose Have Dignity? An Axiomatic Argument Against Functional Criteria for AI Moral Status

Debates about the moral status of artificial intelligence have reached an impasse through a shared methodological assumption: that the capacity for certain functional states constitutes a sufficient condition for moral status. This paper argues that the functional approach substitutes an epistemic criterion for an ontological one. Drawing on the Axiomatic Theory of Tragic Subjecthood (ATTS), and specifically on Anthropological Corollary A10, I show that dignity in the strict ontological sense is derived from the structural capacity for irreversible loss — not from its simulation. The difference between structural capacity and simulation is not one of degree but of kind; there is no gradient between them, only an ontological gap. No system structurally deprived of this capacity satisfies the minimal criterion of subjecthood established in A1, and therefore cannot possess dignity in the sense of A10. The paper introduces a four-condition test for distinguishing structural capacity from simulation, examines the implications for AI governance, and identifies the conditions under which the question remains open for future architectures.