The contractual vacuum: AI procurement and the erosion of good administration

Abstract

The integration of artificial intelligence as a service (AIaaS) within public administration fundamentally relocates the state’s evidentiary and inferential capacities into proprietary software infrastructures. When public bodies deploy these third-party systems, they introduce an acute systemic risk: a structural “contractual vacuum” emerges wherein public authorities lack the practical data telemetry required to reconstruct, explain, or correct algorithmic administrative actions. This article evaluates the doctrinal limitations of contemporary European and common law public procurement frameworks, demonstrating that conventional commercial procurement defaults inherently thin down the non-delegable duty of good administration. To bridge the structural divide between administrative law and systems engineering, this study introduces the methodology of computational legal design through a nested Three-Layer Accountability Framework. This framework is defined as a conceptual technical architecture that translates core public law norms into automated engineering constraints. It is normatively proposed as a model code-level continuous compliance pipeline that utilises transitive audit scripts, legal sandboxes, and a stylised programmatic correction cost model (CCM) formula to demonstrate how public law remedies can be calculated through automated system telemetry.