The artificial intelligence (AI) governance dilemma

The rapid diffusion of Artificial Intelligence (AI) across organizational and societal domains has exposed a tension between innovation, technological capability, and institutional responsibility. While AI systems are often discussed in terms of efficiency, innovation, or risk, less attention has been devoted to the structural misalignment between those who design, control, deploy, and are affected by these systems. This paper develops a theoretically grounded interpretation of AI governance as a problem of distributed agency. Building on Weber, Luhmann, Foucault, and Arendt, it argues that contemporary AI systems cannot be adequately governed through traditional, actor-centric, or Public Administration bureaucracy frameworks. Instead, they require a shift toward system-level accountability. The analysis highlights how current regulatory approaches, like the European Union’s AI Act, only partially address this transformation. The paper concludes by proposing a conceptual model of the ownership–control–responsibility gap and outlining implications for future governance architectures.