Structural Asymmetries in the EU AI Act: A Computational Forensic Analysis of Legislative Architecture

We present a quantitative structural analysis of the European Union Artificial
Intelligence Act using the TRACE Structural Audit Engine, a computational framework for topological and linguistic analysis of legislative texts. Across 113 articles
and 59,306 words of legal body text, we model the Act as a directed network of internal cross-references and measure centrality, bridge influence, vocabulary density,
obligation-rights asymmetry, and term-distribution patterns. The results identify
several measurable structural concentrations: Article 5 functions as the primary
structural anchor of the regulation; Article 74 operates as a disproportionate bridge
node; the enforcement chapters concentrate punitive-coded language; and obligation markers substantially outnumber rights markers. A preliminary comparison
with prior EU digital regulation suggests escalation in obligation density, although
cross-corpus language asymmetry requires further validation. These findings do not
imply legislative intent or normative illegitimacy; rather, they provide a computable
baseline for evaluating the internal architecture of digital regulation.