Canadian AI Register Analysis Reveals Gap Between Sovereign Rhetoric and Bureaucratic Practice

A November 2025 release of Canada's first Federal AI Register prompted critical analysis through the ADMAPS framework. Researchers examined all 409 systems documented, finding that 86% serve internal efficiency purposes rather than public-facing applications. The register's technical focus obscures essential human elements like discretion and training required for system operation. This creates an ontological design that frames AI as purely technical rather than sociotechnical. The paper argues such registers actively shape accountability boundaries rather than neutrally reflecting government activity. Published on arXiv as 2604.15514v1, the study combines quantitative mapping with deductive qualitative coding methods. It reveals systematic omissions in how bureaucratic practice gets represented through transparency mechanisms.