Skip to content
This is NOT an official site of the Government of Canada. Click here for the official AI registry.

Helpful Places

Government of Canada – DTPR AI Register

AI Language Model Fine-Tuned for Canadian Government Context

Inform

What it collects

Operational data
Anonymized data
  • Archival and documentary data from Library and Archives Canada, used to fine-tune the model for Canadian government context. The register states no personal information is involved.

GC AI Register — Data sources field

About behaviour
Anonymized data
  • Queries submitted by GC employees at runtime — the natural language questions or prompts that the model processes to generate responses.
Run by
Shared Services Canada (SSC)
Where
No fixed location
Kept
Not stated by the Helpful Places.
Shared with
Accountable organization
Your copy
You cannot see the data it holds about you. What you can do

What it is for

Inform

GC LLM is a large language model adapted from an open-source base model so that it responds to questions with accurate Canadian government context — for example, knowing that Parliament sits in Ottawa. It is trained using archival data from Library and Archives Canada and runs on Shared Services Canada's high-performance computing infrastructure. The system is intended for use by Government of Canada employees, and users are not currently informed that they are interacting with an AI.

What it collects and what happens to it

Data taken in

Operational data
Anonymized data
  • Archival and documentary data from Library and Archives Canada, used to fine-tune the model for Canadian government context. The register states no personal information is involved.

GC AI Register — Data sources field

About behaviour
Anonymized data
  • Queries submitted by GC employees at runtime — the natural language questions or prompts that the model processes to generate responses.

Processing

Language Models
  • A Meta open-source large language model fine-tuned using data from Library and Archives Canada on SSC's High Performance Computing environment to bias responses toward Canadian government context.

What it does

Understanding (Semantic AI)
Human decides
  • The model reads and understands natural language queries from GC employees and retrieves or applies Canadian government-specific knowledge embedded during fine-tuning.
Creating (Generative AI)
Human decides
  • The fine-tuned LLM generates new text responses grounded in Canadian government context; GC employees receive these responses and decide how to act on them.

Outputs

Generated content
Anonymized data
  • Natural language text responses generated by the fine-tuned model, providing GC-contextualized answers to employee queries. No personal information is produced according to the register.

Run by

Shared Services Canada (SSC)
  • Shared Services Canada (SSC) is the federal department that develops and operates GC LLM, hosting the model on its High Performance Computing environment.

GC AI Register — Department field

Built by

Meta
  • Meta provides the open-source base large language model that Shared Services Canada fine-tunes for Canadian government context.

GC AI Register — Vendor field

Kept for

Not stated by the Helpful Places.

Shared with

Not available to me
  • Output data from GC LLM is not available to the general public. The system is restricted to GC employees.
Available to the accountable organization
  • GC LLM outputs and any interaction logs are available to Shared Services Canada as the operating department.

Stored

Not stated by the Helpful Places.

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 UseThe register explicitly states that AI use is NOT currently disclosed to users. GC employees interacting with GC LLM are not informed they are engaging with an AI system. This is a gap relative to best practice; employees who wish to raise concerns should contact Shared Services Canada.
  • Right to Algorithmic TransparencyGC employees may consult the Government of Canada's public AI register entry for GC LLM (register ID 2526-SSC-SPC-002) to understand the general logic and purpose of the system. Further details about the model architecture, training data selection, and evaluation criteria are not publicly disclosed.

Risks and safeguards

  • Societal & cultural harmA fine-tuned LLM may reproduce biases, inaccuracies, or outdated information from its training corpus, potentially misleading GC employees on policy or factual matters. The model's outputs could also reflect cultural or political framings embedded in Library and Archives Canada data.Safeguard: The model is developed in-house at SSC with domain experts able to evaluate outputs; use is restricted to GC employees rather than the public; the technique was chosen to allow specialization and correction. Further transparency and accuracy-testing measures are not described in the register.
  • Loss of autonomyThe register states that AI use is not disclosed to users, meaning GC employees may not know they are interacting with an AI-generated response, limiting their ability to critically evaluate outputs or seek human alternatives.Safeguard: No disclosure or opt-out mechanism is described in the register. This gap is a significant transparency concern that should be addressed.