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

Back to your search

AI Writing and Research Assistant for Government Employees

Employment & Work

What it collects that can identify you

Sensitive personal information
Identifiable data
  • Employees provide their own text, documents, and queries to the chatbot. This may include personal information about the employee or about third parties referenced in their work, up to and including Protected B data.

Also collects operational data, which is anonymized data.

Run by
Employment and Social Development Canada (ESDC)
Where
No fixed location
Kept
Not stated by the Helpful Places.
Shared with
Accountable organization, Vendor

What it is for

Employment & Work

This system is an AI-powered chatbot available to Government of Canada employees that helps with summarizing documents, editing text, writing code, and conducting research. It uses a large language model (GPT-4o mini) provided by Microsoft and has been approved to handle Protected B information. Employees interact with it voluntarily, and AI use is disclosed to users.

What it collects and what happens to it

Data taken in

Sensitive personal information
Identifiable data
  • Employees provide their own text, documents, and queries to the chatbot. This may include personal information about the employee or about third parties referenced in their work, up to and including Protected B data.
Operational data
Anonymized data
  • Employees may submit operational government documents, code, and research materials as prompts or context for the chatbot to process.

Processing

Language Models
  • EVA Chat is built on the GPT-4o mini large language model supplied by Microsoft. The model processes natural language input and generates text-based responses for summarization, editing, coding, and research tasks.

What it does

Creating (Generative AI)
Human decides
  • EVA Chat generates new text in response to user prompts — drafting summaries, edited documents, code snippets, and research responses. A human employee reviews and decides whether to use any output.
Understanding (Semantic AI)
Human decides
  • The system understands and processes natural language queries submitted by employees, matching intent to relevant responses and retrieving contextual information to support summarization and research tasks.

Outputs

Generated content
Anonymized data
  • The system produces generated text — summaries, edited drafts, code snippets, and research responses — tailored to the employee's query. Output is advisory; employees decide whether and how to use it.

Run by

Employment and Social Development Canada (ESDC)
  • ESDC is the Government of Canada department that deployed EVA Chat for use by federal government employees. ESDC developed the system in part and is accountable for its operation.

ESDC Virtual Assistant (EVA) Chat — Government of Canada AI Register

Built by

Microsoft
  • Microsoft supplies the underlying GPT-4o mini large language model and associated cloud infrastructure used by EVA Chat.

ESDC Virtual Assistant (EVA) Chat — Government of Canada AI Register

Kept for

Not stated by the Helpful Places.

Shared with

Available to the accountable organization
  • Data entered into and outputs produced by EVA Chat are accessible to ESDC as the deploying department for operational and governance purposes.
Available to vendor
  • Microsoft, as the supplier of the GPT-4o mini model and underlying cloud infrastructure, may have access to data processed through the system depending on contractual and data-processing agreements. The extent of vendor access to Protected B content is governed by ESDC's procurement and data handling terms.

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 UseAI use is disclosed to users. Employees are informed that they are interacting with an AI system before and during their use of EVA Chat.
  • Right to Algorithmic TransparencyThe system is listed on the Government of Canada's public AI register, which discloses the model used (GPT-4o mini), the vendor (Microsoft), capabilities, data classification (Protected B), and primary users. Employees may consult the register entry for system-level transparency.

Risks and safeguards

  • Psychological harmEmployees may over-rely on AI-generated content, leading to skill atrophy or uncritical acceptance of inaccurate outputs.Safeguard: AI use is disclosed to users; outputs are advisory only and must be reviewed by the employee before use. The human-in-the-loop design preserves employee judgment as the final step.
  • Civil liberties harmThe system is approved for Protected B data, meaning sensitive personal information about Canadians may be entered by employees. Unauthorized disclosure or mishandling could compromise privacy and due process rights of data subjects.Safeguard: The system is approved only for Protected B classification; ESDC has implemented privacy controls and the system involves personal information safeguards consistent with the Privacy Act of Canada.
  • Societal & cultural harmWidespread use of a single generative AI model across the Government of Canada could homogenize public-service writing, reduce linguistic diversity, and erode institutional knowledge held by employees.Safeguard: The system is positioned as a productivity assistant — outputs require human review and are not intended to replace employee expertise or institutional judgment.