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-Assisted Drafting of Ministerial Correspondence

Inform

What it collects

Operational data
Anonymized data
  • Incoming ministerial correspondence — letters and requests addressed to the Minister — served as the primary input. The register indicates no personal information was involved, suggesting correspondence was treated as operational content rather than personal data.
Run by
Canadian Heritage (PCH)
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

This system used a generative AI model (powered by OpenAI) to draft responses to ministerial correspondence on behalf of Canadian Heritage employees. It was piloted as an experiment to test whether AI-generated drafts could match or exceed the quality and efficiency of human-written responses. The system has since been retired. Members of the public who received ministerial responses during the pilot were not informed that AI assisted in drafting them.

What it collects and what happens to it

Data taken in

Operational data
Anonymized data
  • Incoming ministerial correspondence — letters and requests addressed to the Minister — served as the primary input. The register indicates no personal information was involved, suggesting correspondence was treated as operational content rather than personal data.

Processing

Language Models
  • An OpenAI large language model (specific model version not disclosed in the register) was used to read incoming correspondence and generate draft responses in ministerial correspondence style.

What it does

Creating (Generative AI)
Human decides
  • The system generated draft ministerial correspondence responses. Human GC employees reviewed, edited, and approved drafts before sending — the AI produced content; humans decided whether and how to use it.
Understanding (Semantic AI)
Human decides
  • The system read and understood incoming ministerial correspondence (text input) to ground the generation of appropriate draft responses.

Outputs

Generated content
Anonymized data
  • The system produced draft text responses to ministerial correspondence. These drafts were reviewed and potentially sent as official ministerial replies. The register states no personal information was involved in outputs.

Run by

Canadian Heritage (PCH)
  • The Department of Canadian Heritage piloted and operated this generative AI experiment for drafting ministerial correspondence, integrating it into existing correspondence workflows.

GC AI Register — 2526-PCH-001

Built by

OpenAI
  • OpenAI supplied the underlying generative AI language model used to draft ministerial correspondence during this experiment.

GC AI Register — 2526-PCH-001

Kept for

Not stated by the Helpful Places.

Shared with

Not available to me
  • Data produced by the system (draft correspondence) was accessible only to GC employees involved in the correspondence workflow. Members of the public had no access to the AI-generated drafts or underlying system outputs.
Available to the accountable organization
  • Draft correspondence and system outputs were accessible to Canadian Heritage (DCMF) employees participating in the experiment workflow.

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 UseMembers of the public who received ministerial correspondence during this experiment were not informed that AI assisted in drafting their responses. The register explicitly confirms that AI use was not disclosed to users. This right was not fulfilled during the pilot; the system has since been retired.
  • Right to Algorithmic TransparencyInformation about this AI system is now publicly available through the Government of Canada AI Register. The system has been retired. For questions about its use, contact Canadian Heritage through official Government of Canada channels.

Risks and safeguards

  • Reputational harmAI-generated ministerial responses could contain errors, hallucinations, or inappropriate content that damage the reputation of Canadian Heritage or the Minister.Safeguard: The experiment integrated into existing workflows where GC employees reviewed drafts before sending; adherence to Government of Canada policies and privacy standards was required throughout.
  • Loss of autonomyMembers of the public receiving ministerial correspondence were not informed that AI assisted in drafting their responses, limiting their ability to understand and contest the process. The register confirms AI use was not disclosed to affected individuals.Safeguard: The system has been retired; future implementations are recommended to include disclosure and consent mechanisms.