
This article examines how contemporary AI systems unsettle traditional equality law and why remedies limited to ex post adjudication are insufficient. It maps where bias crystallizes into discrimination across the socio-technical flow of problem framing, data and labeling, model targets and features,deployment choices, and feedback loops-and argues for a principle of "technological equality" that embeds ex ante, lifecycle duties of documentation, participation, testing, oversight, and contestability. A comparative analysis of the Council of Europe's CAI, the EU AI Act, Brazil's PL 2338/2023, and Italy's Law 132/2025 highlights convergences on documentation and oversight, but diverging ambitions on participation, impact assessments, and structural change.
This article examines the evolving architecture of conformity assessment under the EU Artificial Intelligence Act (AI Act), focusing on the interaction between regulatory sandboxes, notified bodies, and the formal certification process. It explores how Written Proofs and Exit Reports, developed within sandbox environments, serve as structured documentation bridging experimental testing and market conformity procedures. The analysis highlights the dual nature of these instruments: as technical evidence supporting compliance and tools for regulatory learning that inform institutional guidance and standards development. Particular attention is given to the role of notified bodies, their potential engagement within sandbox testing under Article 31(5), and the risks of overlap or market distortion. The study argues for a balanced integration of sandbox outcomes into conformity workflows, proposing a cooperative model where regulatory experimentation closely interacts with conformity procedures.
This article aims to plant the seed of a possible ecosystem of artificial intelligence, rooted in hermeneutics, beginning from a concrete proof of existence: the ethical design method Endless Tuning and its implemented prototypes. The ethics of artificial intelligence, by recognizing the importance of vulnerability, becomes the key to the relationships within an infrastructure that includes various figures-manager, developer, ethicist, domain expert, etc. - thus orienting its design firstly on a small scale, in the relationship between the tool/agent and the operator; then expanding, acting as the center of gravity of the social context of that same relationship. According to this order, therefore, we shall proceed: ethics in interface design, the experience of the concrete application of Endless Tuning, and the description of its possible social implications. The latter will present a fictional yet realistic example-a case of medical image diagnostics-as both a possible extension and a recognition of a circular ecosystem that already partially exists.
This paper explores whether large language models (LLMs) can sustain stable moral postures or whether their apparent ethical coherence is merely stochastic. Through a series of zero-shot role-playing prompts combining specific topics and simulated personas, this study31 analyzes linguistic patterns using topic modeling, TF-IDF differentiation, log-likelihood (G2), and measures of semantic convergence and entropy. The results show no enduring moral orientation: the models do not "choose", but statistically recombine fragments of moral discourse inherited from their training corpora. What emerges is a stochastic ethics-an ethics without intentionality, coherence, or agency, yet capable of reflecting human moral structures in probabilistic form. Interpreted through the philosophical framework of moral freedom and the infosphere, the study argues that LLMs act not as moral subjects but as amplifiers of human ethical language, redistributing the moral imaginary of the societies that produce and employ them.
The contribution outlines the work carried out and the results achieved within the Transversal Project (TP) of the PNRR FAIR Project. It begins by providing an overall presentation of the project, followed by an in-depth analysis of the specific features of the TP in all its components, namely the research groups from various Italian universities that belong to some of the Spokes. The TP's ethical and legal focus is addressed by these groups through different approaches, all aimed at creating a digital ecosystem that is ethical, legally compliant, and human-centered.