The Resistance of the Face: Levinas, the Pixel, and the Ethics of the Interval

This paper argues that contemporary large language models (LLMs) and AI-mediated interfaces can simulate interpersonal address while structurally evacuating the conditions under which alterity becomes ethically binding. Building on Emmanuel Levinas’s account of the face as an irreducible ethical summons, I contrast the ontology of the face (exposure, vulnerability, mortality) with the ontology of the pixel (surface, substitution, operational immunity). To strengthen the bridge from ethics to law and governance, I integrate a complementary ontological model from Fernández Tamames’s La Causa Primera: the distinction between Causa Primera (origination under instruction-deficit) and Causa Segunda (transmission of determination), the phenomenology of the interval, and the triadic structure of signature (exposure, appropriation, response). The central claim is that responsibility arises not from conversational fluency or informational richness but from an interval of hesitation in which a vulnerable subject must answer. When institutions routinize black-box mediation and treat outputs as if they were authored acts, they risk an organized diffusion of responsibility. The paper concludes by proposing “sovereign signing” as a governance orientation: preserving deliberative time and requiring reflexive human signature for high-stakes AI outputs as a condition of imputation and contestability.