Acceso usuarios
This article examines the anthropocentric principle in artificial intelligence and the legal category of the “reserve of humanity” in the Italian experience, tracing its development from European soft law to positive law, with particular reference to the AI Act and Law no. 132 of 2025, and to the administrative case law of the Council of State, from judgment no. 2270 of 2019 to the most recent rulings of 2025 and 2026. Adopting a comparative method, the article compares the Italian approach with the Brazilian framework currently under development, showing how the two legal orders – one already equipped with national legislation on artificial intelligence, the other currently considering Projeto de Lei no. 2338/2023 – address the same structural problem, namely the delegation of public decision-making to algorithms, through solutions that converge in principle while diverging in their modes of legal construction and in the instruments adopted. The analysis leads to a distinction between the reserve of humanity and human oversight: the former does not merely amount to technical supervision, but expresses a substantive limit on the delegability of decisions affecting fundamental rights. On this basis, the article argues that the category of a “meaningful human decision reserve”, while rooted in the international debate on meaningful human control, develops it in a specifically legal direction, and may constitute a criterion suited to translating the anthropocentric principle into verifiable operational guarantees, shifting the focus from process control to the human attribution of the final decision.
