Fiscal Geometry as a Data Schema for AI Audits: Rendering Institutional Execution on the X–Y Plane

Fiscal Geometry as a Data Schema for AI Audits introduces Fiscal Geometry (FG) as an engineered representation infrastructure for rendering institutional execution in a machine-readable, auditable form. Rather than asking AI systems to “understand” legal or regulatory narratives, FG provides a pre-analytical schema that measures how rules are executed across jurisdictions and over time. The framework decomposes institutional reality into four minimal primitives—Events, Gates, Interfaces, and Endpoints—and maps them onto an X–Y plane that captures cross-border fiscal movement and intergenerational post-tax capacity. Each object is bound to authoritative sources through mandatory evidence anchors, ensuring that every rendered structure remains traceable, reproducible, and independently verifiable. FG produces three standardized audit outputs: an X–Y event map that spatializes execution paths; a node table suitable for filtering and cross-checking; and a neutral readout that summarizes geometric features without evaluative language. The central contribution is a “pre-analysis layer” for AI-assisted audits. FG does not replace human judgment or infer intent; it stabilizes visibility upstream so that responsibility, interpretation, and sign-off remain human and role-bound downstream. This design enables interoperability and “Bring-Your-Own-Auditor” verification across models, vendors, and institutions. This record establishes FG as a protocol-grade schema with a persistent DOI, suitable for research, pilot implementations, and standards-aligned audit infrastructures across domains such as taxation, compliance, finance, and governance.