Bureaucratic Silences: What the Canadian AI Register Reveals, Omits, and Obscures

Dimensions overall score 3.0

• institutional logics. 4 Results We organize our results through the ADMAPS framework, exam- ining (1) how algorithmic decision-making is represented and op- erationalized; (2) how AI systems are described in relation to bu- reaucratic processes; and (3) how human discretion, expertise, and heuristics are reconfiguredclaim 82%
• has been adopted or adapted by several other departments (e.g., Indigenous Services, Services Canada, Library and Archives Canada, Shared Services Canada), while similar NLP and LLM based tools enable users to“chat with"documents or generate transcripts and summaries from audioclaim 82%
• •Social and professional topics →Governmental regulations; •Applied computing → E-government;•Human-centered computing →Empirical studies in collaborative and social com- puting. Keywordsclaim 82%
• ©2026 Copyright held by the owner/author(s). https://doi.org/XXXXXXX.XXXXXXX Obscures. InProceedings of the 2026 ACM Conference on Fairness, Account- ability, and Transparency (FAccT ’26), June 25–28, 2026, Montréal, Canada.claim 82%
• tation standards, audits, and reporting requirements, as a neces- sary condition for oversight [ 91]. While these approaches have advanced important norms (e.g., datasheet, model cards) [42, 72], critical scholarship [112] has shown that institutional factors (e.g., incentivesclaim 82%
• gration, Refugees, and Citizenship Services Canada (IRCC), rely on highly heterogeneous, context-rich inputs that encode emotional, linguistic, and cultural variation. When optimized for structured analysis, AI systems may flatten ambiguity, overfit superficial pat- terns, and reduce interpretive flexibilityclaim 82%