AI-Assisted Transcription of Historical Canadian Records

Accessibility · Research & Development

What it collects that can identify you

Sensitive personal information

Identifiable data

The source historical records contain personal information, including names and other identifying details of individuals mentioned in newspaper indexes and government orders-in-council dating from 1925 to 1940.

Also collects about a measurement, which is anonymized data.

Run by: Library and Archives Canada (LAC)
Where: No fixed location
Kept: Retained Not specified
Shared with: Accountable organization, Me

What it is for

Accessibility

Research & Development

This pilot project used the Transkribus AI model to automatically transcribe handwritten and printed text from two historical Library and Archives Canada collections: the Index of Canadian Newspapers and Order-in-Council Registers from 1925 to 1940. The goal was to improve public access to these records through digitization and searchable text. Because the source documents contain names and other identifying details, personal information is involved. The system has since been retired.

What it collects and what happens to it

Data taken in

About a measurement

Anonymized data

Scanned images of historical archival documents: the Index of Canadian Newspapers (approximately 117,000 records) and Order-in-Council Registers (1925 to 1940), containing handwritten and printed text.

Sensitive personal information

Identifiable data

The source historical records contain personal information, including names and other identifying details of individuals mentioned in newspaper indexes and government orders-in-council dating from 1925 to 1940.

Processing

Computer Vision

Transkribus uses computer vision to interpret scanned document images, detecting and segmenting text regions before applying Intelligent Character Recognition (ICR) and Handwritten Text Recognition (HTR) models.

Classification & Prediction

The HTR and ICR models classify each image region as a specific character or word, producing structured text transcriptions from historical handwritten and printed document images.

What it does

Sensing (Perceptive AI)

Human decides

The Transkribus AI model senses and interprets handwritten and printed text in scanned document images (ICR and HTR), converting raw image signals into structured textual transcriptions for archival use.

Deciding (Analytical AI)

Human decides

The system classifies and segments characters and words from document images to produce structured transcription outputs. Human archivists review and validate the transcription results.

Outputs

Generated content

Anonymized data

Machine-generated textual transcriptions of the historical archival documents, enabling digitized and searchable access to approximately 117,000 newspaper index records and the Order-in-Council Registers from 1925 to 1940.

Sensitive personal information

Identifiable data

The transcribed output reproduces personal information from the source documents, including names and identifying details of individuals mentioned in historical newspaper indexes and government orders-in-council.

Run by

Library and Archives Canada (LAC)

Library and Archives Canada (LAC) is the federal institution that deployed and managed this pilot transcription project, collaborating with READ-COOP to improve access to historical Canadian records.

Transkribus Transcription — Government of Canada AI Register

Built by

READ-COOP

READ-COOP is the vendor that developed and supplies the Transkribus AI platform, originating from the EU-funded READ project at the University of Innsbruck. It provides the Intelligent Character Recognition (ICR) and Handwritten Text Recognition (HTR) models used in this project.

Transkribus Transcription — Government of Canada AI Register

Kept for

Retained Not specified

The register does not specify a retention period for transcription outputs. As archival records of Library and Archives Canada, the transcribed outputs are likely retained indefinitely as part of Canada's permanent documentary heritage, but no explicit policy was stated.
Duration: Not specified

Shared with

Available to the accountable organization

Transcription outputs and project data are available to Library and Archives Canada employees for archival and access improvement purposes.

Available to me

The transcribed records are intended to be accessible to the public, improving access to historical Canadian archival material.

Stored

Stored locally

As a Government of Canada federal institution project, data is assumed to be stored within Canadian jurisdiction under LAC's custody, though the register does not explicitly confirm storage location or duration.
Duration: Not specified

How to read the colours

Can it identify you?

Anonymized data: Data about people with the link to who is broken. Stripped of identifiers, blurred, aggregated, or noised so this system can’t reasonably tie a record back to an individual.
Pseudonymous data: Each person’s data is tied to a token (hash, ID, template) that lets this system recognise the same person across events, but the token itself doesn’t reveal a name. Reidentification is possible with extra information.
Identifiable data: The data either contains a direct identifier (name, address, account name, recognisable face or voice, plate number) or carries a token this system uses to look up legal identity during processing.

Who completes the loop?

Human decides: This mode suggests; a person decides what to do next. The AI is always advisory — a human is in the loop on every decision. Example: a triage tool ranks cases for a clinician who chooses which to see first.
Human executes: This mode decides; a person carries out the result. Example: an optimizer plans the day’s trash-collection routes, and drivers run them.
Autonomous: This mode decides and acts on its own. No person reviews each decision or carries out the resulting action.

Definitions from the DTPR standard. Amber is about your data, violet about who decides. The fuller the shape and the deeper the colour, the more identifying the data or the less a person is involved.

What you can do

Ask about this system

Questions go to the Helpful Places, not the vendor.

Your rights

Right to Be Informed of AI UseIndividuals whose personal information appears in the historical records have the right to know that AI has been used to transcribe those records. Library and Archives Canada discloses this use through its AI register entry and project documentation.
Right to Algorithmic TransparencyThe public can learn how the Transkribus AI model works — using Intelligent Character Recognition (ICR) and Handwritten Text Recognition (HTR) — through the Government of Canada's AI register and LAC project information.

Risks and safeguards

Reputational harmTranscription errors by the AI model could mis-transcribe names or details of historical individuals, creating inaccurate or misleading records.Safeguard: the project is a pilot with human review expected for archival quality assurance, and outputs are of historical rather than decision-making nature.