Jose Fernández Tamames

This article develops a critical genealogy of legal imputation to show that contemporary debates on AI-assisted legality reactivate a foundational question that modern legal theory tended to leave implicit: the ontological status of the imputable subject. While Kelsen’s formal theory relocates validity to a closed hierarchy of norms and Hart’s institutionalism grounds it in social recognition, both approaches ultimately presuppose a being capable of responding to norms—without fully theorizing that presupposition. The increasing use of algorithmic systems in high-impact legal contexts makes this dependence visible by producing decisions with normative effects while displacing the very locus of accountability. In response, the paper introduces Original Responsibility as the pre-juridical condition that makes imputation possible: the capacity to recognize an act as one’s own, to articulate reasons within a shared horizon of meaning, and to assume consequences. Building on the Kantian background of imputation and Wittgenstein’s rule-following considerations, the article defends an ontological criterion of validity: a norm is fully juridical only insofar as it can be addressed to an entity capable of sustaining its sense as an obligation. The paper concludes that algorithmic systems may conform to rules, but they cannot follow them as norms, and that preserving legal meaning in the algorithmic era requires re-centering responsibility rather than optimizing prediction.

This paper argues that contemporary AI-driven decision systems intensify a form of algorithmic consequentialism in which legitimacy is reduced to predictive performance, while the justificatory structure of judgment is progressively displaced. Against functionalist accounts that treat morally relevant effects as sufficient for moral agency, the paper defends a stricter thesis: normative authority requires ontological vulnerability—the capacity to appropriate an act as one’s own, to answer for it through reasons, and to assume its consequences as a price of freedom. Because current AI systems operate through correlation without semantic understanding, they can generate outputs but cannot sign them in the existential sense that grounds accountability. The analysis distinguishes human reason as a “glass box” (opaque in mechanism yet traceable in justification) from algorithmic “black boxes” whose opacity is constitutive rather than merely technical. To translate this diagnosis into institutional safeguards, the paper introduces the Λ (Lambda) Operator as a binary criterion of reflexive appropriation capacity and proposes a Proof of Signature (PoS) protocol requiring that any high-impact decision assisted by AI be regenerated, appropriated, and explicitly signed by a human agent with Λ = 1. The COMPAS/State v. Loomis line of cases is used to illustrate how explainability tools cannot substitute for imputability, and why a right to explanation is ultimately a right to a responsible signer.

Formal models of agency represent freedom as choice: a selection function over a predefined option set, with rational agency the optimization of that selection. This paper argues that the representation misses the cases that matter most. Founding a legal institution, inventing a technique, making a work of art — paradigm exercises of freedom — do not select from a given set of options; they enlarge the set, introducing possibilities not deducible from prior rules or antecedent reasons. The paper formalizes this as poietic freedom and defines it through the Free* operator, a construction satisfying five conditions that jointly separate option-set expansion from selection within a fixed set. The central result is a formal, not merely rhetorical, distinction: optimization, however unbounded its search, operates over a domain it cannot itself extend, whereas poietic acts alter the domain. The thesis is bounded to the structure of the operation and takes no stand on metaphysical libertarianism; it shows only that a wide and important class of free acts is formally irreducible to constrained maximization. The payoff is a precise criterion for when an agent — human or artificial — is exercising freedom in the poietic sense rather than performing sophisticated selection.

¿Puede el ser humano seguir siendo el origen en un mundo donde la máquina es el resultado perfecto? Vivimos bajo el asedio del Nuevo Barroco Tecnológico: un ecosistema de ruido infinito, plausibilidad vacía y algoritmos que no solo predicen nuestras respuestas, sino que las anulan antes de que nazcan. En este escenario, la "humanidad" ya no se defiende con el sentimiento, sino con la estructura. La Causa Primera no es un ensayo sobre inteligencia artificial; es un tratado de ontología de urgencia. Lejos del optimismo ingenuo de Silicon Valley y del catastrofismo romántico europeo, esta obra establece una distinción radical que lo cambia todo: La Causa Segunda (La Máquina): La perfección operativa, la optimización estadística y la administración de lo probable. La Causa Primera (El Humano): La capacidad de interrumpir la inercia, asumir el riesgo de la "Verdad Rota" y firmar un acto no garantizado por la estadística. Estructurado como una Sinfonía filosófica en cinco movimientos con obertura, este libro desmonta la falsa equivalencia entre procesar datos y pensar. Desde la deconstrucción de la "Caja Negra" jurídica hasta la formulación de la "Indigencia Biológica" como ventaja operativa, el autor demuestra por qué nuestra vulnerabilidad es la única tecnología que la IA no puede replicar. Esto no es un libro para ser leído; es un manual para dejar de ser un usuario y volver a ser un Arquitecto. La máquina calcula. Tú sostienes el riesgo. Tú eres la Causa Primera.

Large language models and AI-mediated interfaces now simulate interpersonal address convincingly enough to elicit the responses a human face elicits. This paper asks whether such address can be ethically binding, and argues that it cannot. Building on Levinas's account of the face as an ethical summons, it isolates three conditions under which address obligates its recipient: exposure to harm, mortality, and non-substitutability — the one who addresses stands as singular and cannot be swapped without remainder. Mediated interfaces meet none of the three: they are substitutable by construction, immune to harm, and outside mortality. The paper formulates this as the difference between the face and the pixel, and converts the contrast from metaphor into criterion: a source of address binds only if withdrawing from it costs the source something it cannot recover. On that test, simulated address is not weakly binding but structurally non-binding, however sincere the user's response. The argument is bounded — it concerns the conditions of obligation, not the phenomenology of the encounter — and it carries a legal corollary: an interface that cannot be summoned cannot be answerable, which places the duty on the human who deploys it, not on the interface that addresses us. The face resists because it can be lost; the pixel cannot, and so cannot resist.

Modern legal orders presuppose that acts have authors: someone can be asked “why this?” and someone can answer. This paper argues that algorithmic decision-making destabilizes that institutional grammar of imputation in two coupled ways. First, it erodes agency in practice through routine delegation, converting professionals into relays of system outputs. Second, a technical shift from deterministic, inspectable procedures (“glass box”) to probabilistic, opaque learning systems (“black box”) converts legitimacy from reason-giving to performance, producing decisions that may be effective yet structurally resistant to attribution. Methodologically, the paper combines conceptual analysis, doctrinal reading of EU governance instruments (GDPR and the EU AI Act), and institutional analysis of accountability pipelines. Drawing on the triadic model of agency (initiate, judge, respond) and the ontology of signature developed in LA CAUSA PRIMERA (Fernández Tamames, 2026), it argues that explainability, while valuable, is insufficient as a remedy, and that “electronic personhood” is a liability-displacement fiction. The political remedy proposed is sovereign signing: institutionalized, reflexive human signature for high-stakes algorithmic outputs, coupled with contestability, traceability, and enforceable liability anchoring. Imputation is thereby defended as political resistance against organized irresponsibility in algorithmic governance.

Contemporary discourse on Artificial Intelligence predominantly frames human agency as a psychological attribute or ethical imperative. This paper argues for a structural redefinition grounding freedom in ontology rather than ethics. We establish that human freedom is not a consequence of rationality but its precondition: humans are free (ontologically incomplete), therefore rational. This inverts 2,500 years of Aristotelian tradition. Ontological Indigence—the biological reality of neoteny and the absence of closed instinctive programs—is not the source of freedom but its observable manifestation at the biological level. By distinguishing First Cause (capacity to initiate causal chains not determined by antecedent conditions) from Second Cause (transmission of received determinations), we analyze the specific threat posed by Generative AI. While AI systems simulate creativity, they operate structurally as perfect Second Causes: processing syntax without semantic comprehension or the biological necessity of the "Interval" between stimulus and response. The paper concludes that human freedom is an ontological structure manifested as biological indigence, and that the automation of judgment threatens to erode the cognitive structures required to exercise First Cause, effectively reducing human agents to operators of statistical probability.