Public-Facing AI Chatbot Governance

University Relations provides governance and approval for AI-powered chatbots and conversational experiences presented to external audiences under the Queen’s University brand.

This governance is sponsored by the Associate Vice-Principal, Marketing and Brand and the Associate Vice-Principal, Integrated Communications, reflecting University Relations’ accountability for Queen’s public-facing brand, marketing, and communications.

This governance is intended to help units use AI responsibly while protecting the quality, consistency, accessibility, and reputation of Queen’s public-facing communications.

What requires review and approval

Queen’s units planning to launch a public-facing AI chatbot, assistant, or conversational interface must receive approval from University Relations before it is made available to external audiences.

Examples may include:

• AI chatbots on Queen’s websites
• Embedded conversational assistants
• AI-powered conversational search experiences
• Voice or messaging-based AI experiences
• Other AI interfaces that communicate with external audiences on behalf of Queen’s

This requirement applies whether the solution is developed internally or provided by an external vendor.

What University Relations reviews

University Relations reviews public-facing AI experiences from a communications and institutional reputation perspective, including:

• Brand alignment
• Communications quality and accuracy
• Voice and tone
• User experience
• Accessibility
• Transparency and appropriate disclosure of AI use
• Content governance and maintenance
• Reputation risk
• Appropriate testing and quality assurance before launch

The intent is to keep the governance process practical and proportionate so that units can adopt useful AI capabilities without unnecessary administrative burden.

Testing and quality assurance

Units seeking approval are responsible for ensuring that their proposed public-facing AI experience has a testing mechanism capable of receiving and running test questions or scenarios provided by University Relations. This capability must be in place before approval testing begins and may be required for ongoing testing after launch.

University Relations will maintain a dynamic and evolving test bank and will provide the questions or scenarios to be tested. The unit, service owner, vendor, or platform team is responsible for configuring and operating the testing mechanism, running the tests, and providing the resulting outputs to University Relations for review. University Relations will not configure or execute testing on behalf of units.

Where appropriate, units may use Queen’s enterprise chatbot tools with built-in testing capabilities, or another mechanism that can reliably run the University Relations test set and produce reviewable results.

Testing may assess areas such as:

• Accuracy and response quality
• Brand, voice, and communications standards
• Accessibility and user experience
• Transparency and disclosure
• Inappropriate, unsafe, or misleading responses
• Knowledge source integrity
• Reputation and governance risks

Automated testing is intended to support consistent and scalable quality assurance. University Relations will review the test results and determine whether further testing, remediation, or human review is required before approval.

Other institutional requirements

University Relations governance does not replace other institutional requirements that may apply to an AI implementation.

Depending on the technology, data, and use case, additional review or involvement may be required in areas such as:

• Information security
• Privacy
• Data governance
• Technology architecture and integration
• Authentication and access
• Procurement or vendor management

When these additional areas need to be considered, University Relations will coordinate with Information Technology Services and other institutional partners. 

How to engage University Relations

Units are encouraged to contact University Relations early when considering a public-facing AI chatbot. Early consultation can help identify potential issues before significant development or procurement decisions are made.

The review process is structured to be swift and efficient:

1. Share the proposed use case and intended audience.
2. Identify the technology or platform being considered.
3. Provide information on the content, data sources, and user experience.
4. Confirm a testing mechanism that can run University Relations-provided questions or scenarios; the unit is responsible for executing the tests and providing the results to University Relations for review.
5. University Relations reviews the proposed experience and identifies any required changes or additional institutional reviews.
6. University Relations provides approval for the public-facing communications experience before launch.

More detailed guidance, review criteria, and supporting resources are currently being developed and will be published as they become available.

Scope

This page describes University Relations governance for public-facing AI communication experiences.

It does not establish governance for AI used primarily for:

• Teaching and learning
• Research
• Internal productivity
• Internal operational workflows
• Employee-only or other internal AI tools

Those uses may be subject to other Queen’s policies, standards, or institutional AI governance processes.

Questions about whether a proposed AI experience falls within this scope can be directed to University Relations.