The federal government launched its comprehensive Artificial Intelligence Public Registry yesterday, cataloging more than 400 AI systems currently deployed or under development across various departments. This transparency initiative makes Canada one of the few countries globally to publicly document government AI use at this scale.
“This registry isn’t just about checking a regulatory box,” said Minister of Innovation, Science and Industry François-Philippe Champagne at the virtual launch event. “It’s about building public trust through transparency as AI becomes increasingly embedded in government operations.”
The registry, accessible through a searchable online portal, reveals the surprisingly wide reach of AI adoption across federal services. Projects range from predictive algorithms in border security to natural language processing tools that summarize ministerial briefings. Each entry includes information about the system’s purpose, how it works, risk assessment results, and which department oversees it.
What’s striking about the registry isn’t just its size but the diversity of applications. The Canada Revenue Agency leads with 67 AI implementations, primarily focused on tax fraud detection and automated assessment tools. Meanwhile, Immigration, Refugees and Citizenship Canada has deployed 52 systems, including controversial predictive models for visa processing and settlement pattern analysis.
“We’re seeing a government that’s quietly transformed itself into a significant AI operator,” said Michael Karlin, executive director of the AI Governance and Oversight Office. “These systems touch Canadians’ lives in ways most people don’t realize.”
The registry emerges amidst growing global debate about algorithmic accountability in government. Several Canadian civil liberties organizations have already begun analyzing the registry data, with initial reactions mixing cautious approval for transparency with concerns about implementation.
“Finally having this information public is a crucial first step,” said Maya Kotecha, director of digital rights at the Canadian Civil Liberties Association. “But what we’re seeing raises serious questions about algorithmic impact assessments and whether these systems are being deployed with adequate safeguards.”
A deeper look at the registry reveals that approximately 40% of the systems are classified as having “moderate to high” impact on rights, health, economic interests, or the environment. These higher-risk systems face additional governance requirements, including human oversight mechanisms and regular auditing.
One such system deployed by Employment and Social Development Canada uses predictive modeling to flag potential Employment Insurance fraud, resulting in approximately 14,000 claims being flagged for human review monthly. The registry entry notes that the system underwent three rounds of bias testing before deployment.
Industry observers note that the registry also functions as a revealing snapshot of Canada’s AI procurement patterns. IBM, Microsoft, and Amazon emerge as the dominant vendors, though several Canadian companies including Element AI (now part of ServiceNow) and Mindbridge Analytics also feature prominently.
“This shows the government has become a major client for AI services,” explained Raquel Urtasun, professor at the University of Toronto and founder of self-driving technology company Waabi. “That creates both opportunities and responsibilities for the Canadian AI ecosystem.”
Privacy Commissioner Philippe Dufresne expressed qualified support for the registry while noting his office will monitor how listed systems align with privacy law. “Transparency is necessary but not sufficient,” Dufresne said in a statement. “We’ll be watching closely how these systems handle personal information and whether Canadians are adequately informed when AI systems make decisions affecting them.”
The registry also reveals surprising gaps. National security agencies are exempted from full disclosure requirements, though they must still document systems internally. Additionally, some departments appear underrepresented despite their size and public-facing responsibilities.
The timing of the registry’s launch coincides with Canada’s implementation of the Artificial Intelligence and Data Act (AIDA), which establishes new regulatory requirements for high-impact AI systems both in government and the private sector.
For ordinary Canadians, the registry offers unprecedented visibility into how algorithms influence government services they interact with. A Toronto resident checking their tax assessment, a newcomer applying for permanent residency, or a small business seeking government funding may now trace which automated systems influenced those processes.
“We’ve entered an era where algorithmic decision-making touches nearly every aspect of governance,” said Teresa Scassa, Canada Research Chair in Information Law and Policy at the University of Ottawa. “This registry helps democratize that knowledge.”
Going forward, the registry will update quarterly, with all federal departments required to document new AI implementations within 60 days of deployment. The government has also launched a feedback mechanism allowing the public to report concerns about specific systems.
As I scrolled through the registry entries yesterday, what struck me wasn’t just the scale of AI adoption but how quietly it had happened. Many of these systems have operated for years without public knowledge. Whether this new transparency leads to meaningful accountability remains to be seen, but at minimum, Canadians now have unprecedented visibility into how their government is being transformed by algorithms.
