The Government of Canada (GC) Artificial Intelligence (AI) Register collects information about the Government of Canada’s AI systems. Its publication fulfils a commitment made in the AI Strategy for the Federal Public Service 2025-2027.
This version of the GC AI Register is called a minimum viable product (MVP), an early version with only basic features and content. It collects information about AI systems that are or have been in use within the GC.
The register shows that LAC has engaged in 8 distinct AI initiatives from 2018 to 2025. The majority of these initiatives focus on digitizing, classifying, and transcribing historical records to make them searchable for the public and government employees. Six entries are completed or “Retired” pilot projects; two major systems remain “In Production.” One of those two, the 1931 Census, is classified as for public access. It is unclear which of the “Retired”, legacy projects have resulted in publicly accessible material.
Key Areas of Focus
1. Historical Transcription & Public Access (Genealogy/Research)
LAC utilizes AI (specifically OCR and Handwritten Text Recognition) to convert scanned images into searchable text.
-
1931 Census Index: A production-level partnership with Ancestry and FamilySearch to index ~10 million names.
-
Indigenous Records (RG10): A massive project to transcribe ~6 million pages of the Department of Indian Affairs records using Transkribus.
-
Newspapers & Orders-in-Council: A pilot using Transkribus to transcribe Canadian newspapers (high success rate) and legal registers (mixed success due to scan quality).
-
Immigration Records: Used AWS Textract to extract data from over 11,000 pages of the Canada Gazette for genealogy search.
2. Internal Data Management & Classification
Several pilot projects were aimed at organizing internal government data and reducing “ROT” (Redundant, Obsolete, Trivial) information.
-
Network Drive Cleanup: A project with Oproma ( project management software) that successfully classified 3.5 million documents and moved them to GCdocs.
-
Metadata Enrichment: Pilots to automatically categorize archival files and generate metadata for government publications to improve information architecture.
3. Generative AI & Sovereignty
-
CANChat: A current production system developed with Shared Services Canada. It is a generative AI chatbot trained on 50 TB of Canadian sovereign data (including the Government of Canada Web Archive) to assist employees with drafting and research while ensuring data remains in Canada. See a bit more detail at https://www.canada.ca/en/shared-services/campaigns/stories/canchat-sscs-first-generative-ai-chatbot.html
Project Status Summary