Canada's Federal AI Register — 409 Systems Disclosed, but Only 4% With Public Impact Assessments

On 28 November 2025, Canada's Treasury Board published the Government of Canada AI Register, the country's first public inventory of federal AI use. Announced by Treasury Board President Shafqat Ali, the register draws on data validated by departments and lists 409 AI systems — ranging from early research and proof-of-concept work to fully deployed operational tools — across 42 federal institutions.
Each entry records a system's purpose, intended use, and whether it was built in-house or by a vendor; a companion open-data release lets researchers analyse the register directly. The government describes this first release as a minimum viable product, with public consultations planned for 2026 to refine what gets published and how.
An independent academic analysis published shortly after launch found the register's coverage was uneven: only 4% of 303 automated tools it examined had a publicly available Algorithmic Impact Assessment, despite that being formally required under Canada's 2019 Directive on Automated Decision-Making. The same analysis found 86% of listed systems serve internal government operations rather than citizen-facing services, and flagged the omission of at least one high-scrutiny system — Immigration, Refugees and Citizenship Canada's Chinook visa-processing tool. A single vendor, Microsoft, accounts for 46 of the listed systems.
Read together, the register and its independent critique illustrate both the value and the limits of self-reported federal AI inventories: 409 systems is a genuinely large disclosure by international standards, but the gap between the register's existence and the underlying Directive's impact-assessment requirement shows disclosure alone does not guarantee accountability.

Read the full analysis: https://www.theglobeandmail.com/investing/markets/markets-news/Newswire.ca/36361700/canada-launches-first-register-of-ai-uses-in-federal-government/