Responsabilità algoritmica e architettura giuridica : Verso una governance distribuita del rischio

The algorithmic revolution has disrupted the foundational pillars of tradi- tional legal responsibility. Artificial intelligence systems – especially those based on deep learning and opaque architectures – operate without moral agency or intentionality, yet they produce both normative and material effects. This essay investigates the crisis of the classical paradigm of liability – grounded in linear causality, subjective fault, and imputability – and proposes a systemic reconfigu- ration based on the environmental distribution of risk. The approach embraces a relational and architectural perspective: responsibility is no longer conceived as a retrospective attribution, but as an anticipatory and distributed normative func- tion embedded in technical-legal contexts. Drawing on Floridi’s theory of the infosphere and moral patients, Santoni de Sio’s concept of traceable responsibil- ity, Porter’s ecological accountability, and Stahl’s notion of meta-responsibility, the paper develops a model of design-based responsibility, entrusted to human agents who co-construct digital environments and act as systemic custodians of risk. The contribution also offers a critical reading of Ruffolo’s objectivizing ap- proach, advocating instead for a multilayered, preventive, and governance-ori- ented framework for computational harms. The analysis combines theoretical insight with European legal developments.