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

On-Premises AI Research and Experimentation Framework

Research & Development

What it collects that can identify you

Sensitive personal information
Identifiable data
  • The system involves personal information, which may be present in certain internal datasets or documents processed during R&D experiments; the register confirms personal information is involved.

Also collects operational data, which is anonymized data.

Run by
Public Service Commission of Canada (PSC)
Where
No fixed location
Kept
Not stated by the Helpful Places.
Shared with
Accountable organization, Not available to vendor

What it is for

Research & Development

This system provides Public Service Commission of Canada developers and data specialists with a secure, on-premises environment to experiment with large language model (LLM) technology. It uses the open-source Ollama framework to run AI models locally, ensuring that internal data is never sent outside the organization. The system is currently in development and is used for research and exploratory work rather than operational decision-making affecting the public.

What it collects and what happens to it

Data taken in

Operational data
Anonymized data
  • Various internal datasets and documents used according to effort-specific R&D requirements. These are organizational records and documents, not data tied to specific individuals' interactions with a public-facing service.
Sensitive personal information
Identifiable data
  • The system involves personal information, which may be present in certain internal datasets or documents processed during R&D experiments; the register confirms personal information is involved.

Processing

Language Models
  • The Ollama framework hosts a variety of open-source large language models locally on PSC infrastructure, enabling text understanding and generation tasks in a Protected environment.

What it does

Deciding (Analytical AI)
Human decides
  • Researchers and developers use the framework to run analytical experiments on internal datasets, with all outputs reviewed and acted upon by human specialists rather than triggering automated decisions.
Creating (Generative AI)
Human decides
  • The hosted LLMs can generate text outputs as part of research and experimentation; all generated content is reviewed by human specialists before any downstream use.

Outputs

Generated content
Anonymized data
  • The system produces LLM-generated text outputs as part of research and experimentation. Outputs are used internally by PSC specialists to evaluate the utility of LLM approaches for potential future solutions.
Operational data
Anonymized data
  • Experimental outputs such as analysis results, summaries, and model evaluations generated during R&D are retained for internal use by PSC developers and data specialists.

Run by

Public Service Commission of Canada (PSC)
  • The Public Service Commission of Canada is the federal government department accountable for deploying and operating this self-hosted LLM framework for internal research and development purposes.

Government of Canada AI and Data Solutions Register — 2526-PSC-CFP-001

Built by

Ollama
  • Ollama is the open-source framework that provides the local LLM hosting capability used by this system. As an open-source project, it has no single commercial vendor; the PSC selected it for its ability to run powerful AI models on-premises.

Government of Canada AI and Data Solutions Register — 2526-PSC-CFP-001

Kept for

Not stated by the Helpful Places.

Shared with

Available to the accountable organization
  • Outputs and data processed by the system are available only to authorized PSC employees (developers and data specialists) within the organization's Protected on-premises environment.
Not available to vendor
  • Because Ollama is deployed self-hosted on PSC infrastructure, no data is sent to Ollama or any external party. The vendor has no access to data processed by this system.

Stored

Stored locally
  • All data is processed and stored on PSC on-premises infrastructure within Canada, in a Protected environment, consistent with the self-hosted design of the system.
  • Duration: Not specified in the register
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 confirms that AI use is disclosed to users of this system. GC employees using the framework are informed that they are working with an AI-powered tool.
  • Right to Purpose LimitationThe system is explicitly scoped to experimentation, research and development work. Data processed within the framework is not to be used for operational decisions affecting individuals; the on-premises design ensures data stays within the organization.

Risks and safeguards

  • Civil liberties harmInternal datasets may include personal information about individuals, and LLM experimentation on such data could expose sensitive content or generate outputs that improperly characterize individuals.Safeguard: The system operates in a Protected on-premises environment, ensuring data never leaves the organization; access is limited to PSC developers and data specialists; the system is in development with no operational decisions made by the AI.