TRU horaizon

TRU AI Tools Guidance

Purpose

This webpage provides clear, practical guidance for the TRU community on the appropriate use of AI tools.

It is intended to:

  • Guide safe and responsible use of AI, not to grant approval or authorization.
  • Avoid endorsing non-approved technologies.
  • Align with TRU procurement, legal, risk, and privacy (PIA) processes.
  • Complement institutionally curated learning technologies, including AI enabled tools supported through Learning Technology & Innovation (LT&I).

It does not replace or bypass:

  • Formal procurement and contracting processes.
  • Required risk and privacy assessments.
  • Existing academic, administrative, or IT governance policies.

Core Principle

The risk of AI use is determined by the sensitivity of the data, not by how powerful or popular a tool may be. 

Institutional AI Access Model

Copilot-First AI Usage Standard

To ensure secure, compliant, and efficient adoption of artificial intelligence tools, Thompson Rivers University (TRU) follows a Copilot-First AI Usage Standard. This approach prioritizes institutionally approved tools operating within secure enterprise environments before considering external AI platforms. 

Learn more about Copilot, approved use cases, and default requirements

Layered AI access model

TRU supports academic success through a layered AI access model, combining institutionally managed tools with carefully assessed external platforms.

Institutional Access

TRU provides secure access to:

  • Microsoft 365 Copilot

Approved for Public, Internal & Confidential data withing M365 environment.

Library Promoted AI Resources

The TRU Library promotes AI-supported research tools: 

  • EBSCO AI Insights 
  • EBSCO Natural Language Search 

These tools support literature discovery and evidence-based academic research. 

Writing & Academic Support Tool

Widely used tools:

  • Grammarly

Requires risk & privacy assessment before broader endorsement. Public or limited Internal data only.

TRU Data Classification

AI usage decisions must be based on TRU’s official Data Classification framework. 

Before using any AI tool, users are required to review the definitions and examples outlined in the HorAIzon Guidelines – TRU Data Classification webpage to determine whether the information involved is classified as Public, Internal, or Confidential. 

The classification of the data — not the tool — determines whether AI use is permitted. If there is uncertainty regarding the classification level, users must consult their supervisor or the appropriate institutional authority before proceeding. 

Failure to align AI usage with TRU Data Classification standards may result in privacy, legal, or regulatory risk. 

PIA Decision Workflow Steps

1

What type of data will you use?

Check the TRU Information Classification page to classify your data:

  • Confidential
  • Internal
  • Public

2

Is the data Confidential or Internal

Yes → PIA required.
No (Public only) → Go to Step 3.

3

Is the tool external or cloud-based?

TRU treats most online tools (including AI tools) as cloud services, which trigger additional privacy obligations under FIPPA.

Ask:

  • Is the data stored or backed up outside Canada? 
  • Does the vendor retain the data?
  • Does the vendor have access to the data? 

If YES → A PIA is required 
If NO → Go to Step 4 

4

Have you verified the privacy, security, and data‑storage practices?

  • Have you reviewed the vendor’s privacy policy and terms of use? 
  • Do you know where the data is stored and processed?
  • Do you know whether the data is retained, reused, or accessed by the vendor? 

If NO or NOT SURE → PIA required 
If YES → You may use the tool without a PIA 

AI Model Training, Data Control, and Opt-in / Opt-out Considerations

When using any AI tool, it is important to understand how your data is handled after it is submitted, including whether it may be retained, stored, or used to train the underlying AI model. 

Some AI vendors allow users to opt out of having their data used for model training, by default the user is always opt-in, or some AI vendors do not offer a choice at all. 

Before using an AI tool, especially for TRU work, users should review the tool’s privacy or data controls and confirm: 

  • Whether submitted data is used for model training
    • Is data used to improve or retrain the model?
    • Is training enabled by default, or only with user consent? 
  • Whether training is opt-in or optout 
    • Optout: Data is used for training unless the user disables it. 
    • Opt-in: Data is not used unless the user explicitly enables it. 
  • Whether data controls are available and enforceable 
    • Can training be disabled at the account or organization level? 
    • Do settings apply consistently across chats, uploads, and prompts? 
  • Where this information is documented
    • Vendor privacy policy 
    • Terms of service 
    • Data controls or account settings pages 

Example: In AI vendors like ChatGPT, Claude, Gemini, users can disable “Improve the model for everyone” under Data Controls, which prevents submitted data from being used for model training.  

  • Be retained beyond TRU’s control 
  • Be stored or processed outside Canada 
  • Be reused in ways that are not transparent or reversible 
  • Create privacy, legal, or reputational risks if the data includes personal, internal, or confidential information 

For these reasons:

  • If an AI tool does not clearly allow users to opt out of data being used for training, it must not be used with Internal or Confidential data. 
  • If you cannot determine how data is stored, retained, or reused, please consult with InfoSecurity@tru.ca. 
  • If optout controls are unclear, unavailable, or cannot be verified, the tool should be treated as higher risk. 

If the data is Confidential or Internal → submit a PIA. 
If the data is Public only → a PIA is not required. 

  • Need help deciding? Review the resources below to understand how TRU defines personal information and when a Privacy Impact Assessment (PIA) is required. 

More Resources Available

Personal Information: Thompson Rivers University

Collecting Personal Information, Thompson Rivers University Privacy Rules

Privacy Rules

Privacy Impact Assessments: Office of the General Counsel and Secretariat: Thompson Rivers University

The tools listed below are provided as examples only. 

Their inclusion does not imply endorsement, recommendation or approval by TRU. 

Except for Microsoft 365 Copilot, these tools must be used with Public data only. 

Tool

Data Type

User Type

Example Use Case

Microsoft Copilot 

Public, Internal, Confidential (secure M365 only) 

Staff & Administration; Faculty; Researchers 

Policy drafting, internal reporting, secure documentation, course material development, research drafting 

ChatGPT  

Public

Students; Faculty; Staff; Researchers 

Academic drafting, brainstorming, study support, teaching preparation

Google Gemini

Public only 

Students; Faculty

Brainstorming, public content drafting, concept exploration

Perplexity AI

Public only 

Students; Faculty; Researchers

Public information search, literature exploration, citation discovery 

Tool

Data Type

User Type

Example Use Case

EBSCO AI Insights 

Public 

Faculty; Researchers; Students (research queries) 

Evidence discovery, literature mapping, topic refinement 

EBSCO Natural Language Search 

Public 

Faculty; Researchers; Students 

Database searching using natural language, academic source discovery 

Scite.ai  

Public research queries only 

Researchers; Faculty 

Citation verification, evidence strength evaluation 

Scopus AI

Public exploration only 

Researchers; Faculty 

Literature review assistance, research trend analysis 

Tool

Data Type

User Type

Example Use Case

Grammarly

Public; limited non-sensitive Internal 

Students; Faculty; Staff 

Grammar correction, writing clarity, style refinement 

Quillbot  

Public only 

Students; Faculty 

Paraphrasing, sentence restructuring 

Wordtune

Public only 

Students; Faculty; Staff 

Tone adjustment, writing refinement 

Tool

Data Type

User Type

Example Use Case

Read.ai 

Public 

Faculty; Staff; Students 

Meeting transcription, email assistant, summaries 

Notion  

Public only 

Students; Staff 

Personal notes, task tracking, public project planning 

Otter.ai  

Public sessions only  

Faculty; Staff; Students 

Meeting transcription, lecture capture, interview notes 

Tool

Data Type

User Type

Example Use Case

GitHub AI (Coding Support) 

Public code; approved Internal development projects 

Researchers; Faculty; Students (programming courses) 

Code completion, debugging, software prototyping 

Julius.ai

Public datasets only 

Researchers; Students 

Data analysis, visualization, dataset exploration 

DALL-E

Public creative assets only 

Students; Faculty; Staff 

Image generation, academic visuals, presentation graphics 

Tool

Data Type

User Type

Example Use Cases 

LLAMA4 (Maverick)

Controlled testing environments only 

Researchers; Institutional AI Teams 

Multimodal experimentation, research prototypes, safeguarded model testing 

Tool / Category 

Status 

Reason 

DeepSeek (all apps) 

Banned 

Severe privacy and security concerns 

Deepfake / faceswap tools 

Banned 

Identity and reputational risk 

Zoom AI Companion 

Restricted 

Consent and recording risk 

Any unapproved AI with Confidential data 

Not permitted 

No contractual or privacy protections 

Consideration

  • Power Automate and Power Apps are approved only when used within TRU’s licensed Microsoft 365 environment and in accordance with TRU Data Classification standards. 

Related Policies & Guidance

This Copilot usage guidance aligns with TRU’s broader institutional records and information management framework. 

For information on institutional records retention, classification, and management responsibilities, see TRU Records Management Documentation.

Learn More