Artificial Intelligence in Administrative Law: Structural Deficits, Control Gaps and Platform Dependency. A Jurisprudential and Administrative Law Analysis

This paper analyses the deployment of artificial intelligence (AI) in public administration across four structurally distinct yet normatively interconnected levels. It connects jurisprudential foundations — Kelsen’s Grundnorm and Hart’s Rule of Recognition — with an administrative law review of the structural mismatches between the historically sedimented Federal Administrative Procedures Act (Verwaltungsverfahrensgesetz, VwVfG) and the logic of probabilistic language models; the European law framework established by the AI Regulation (EU) 2024/1689 and the AI Market Surveillance and Innovation Act (Künstliches-Intelligenz-Marktüberwachungs und Innovationsförderungsgesetz, KI-MIG); the institutional modernisation policy of the Federal Ministry for Digital Affairs and State Modernisation (Bundesministerium für Digitales und Staatsmodernisierung, BMDS); and the geopolitical dimension of digital dependency on externally controlled platform architectures. The central jurisprudential question is: whose normative claim to validity sustains the administrative decision when the underlying AI system is governed by an external, state-uncontrollable platform architecture?