AI Usage Policy

1. Our commitment

Fraser Insights uses artificial intelligence (AI) tools only where they help us do better research, and never at the expense of client confidentiality or the trust of research participants. This policy explains when and how we use AI, and the safeguards that apply.

We follow the CRIC Guiding Principles for AI use in Market Research, including the commitments to transparency, data security, protecting participants from harm, working to minimize bias, and ensuring human oversight. We handle personal information in line with Canada's Personal Information Protection and Electronic Documents Act (PIPEDA), any provincial privacy laws that apply to our work, and the CRIC Code of Conduct and Standards. This policy should be read alongside our Privacy Policy.

2. Accountability and human oversight

Erik Fraser is Fraser Insights' designated Privacy Officer and is responsible for our compliance with this policy. Questions and concerns can be sent to erik@fraserinsights.ca.

AI does not make decisions for us. A researcher reviews, verifies, and takes responsibility for every AI-assisted output, and all final deliverables, including analysis, findings, recommendations, and reports, are created by people. We apply our existing research ethics review to AI-related risks, uses, and safeguards, with the same rigour used for other research methods. Anyone who works with us, including employees, contractors, and interviewers, must follow this policy and sign a confidentiality agreement before accessing client or participant information.

3. How we may use AI

With client agreement, AI tools may be used in a project to assist our researchers. Typical uses include:

  • Questionnaire and guide development: generating draft questions, probes, and response options, and optimizing survey flow and wording.

  • Fieldwork and data preparation: supporting clearly disclosed uses such as AI-enabled fieldwork, data cleaning, or quality checks, subject to researcher oversight.

  • Analysis support: initial analysis of data, such as first-pass coding and theming of open-ended responses and identifying patterns for a researcher to investigate.

  • Transcription and summarization: transcribing qualitative sessions and producing draft summaries for researcher review.

  • Drafting and editing: assisting with early drafts, report generation, formatting, and proofreading of project materials.

  • Synthetic data or personas: using simulated data or AI-generated personas only where appropriate, agreed with the client, clearly identified as synthetic, and never represented as responses from real participants.

  • Where a client prefers that AI not be used on their project, we will respect that, and we will not use AI on the project.

4. What we do not do

  • We do not enter client names, proprietary ideas or materials, or respondent identifiers into AI tools that have not been approved under this policy.

  • We do not allow client or participant data to be used to train or improve third-party AI models.

  • We do not use AI to make decisions about individual participants, or to identify or re-identify people in de-identified data.

  • We do not present AI-generated content, including synthetic or simulated respondents, as real research data or as the views of real people.

  • We do not release AI-generated findings or recommendations without human review and validation against the underlying data.

5. Protecting research participants

  • Transparency and consent: Where AI is used to process participant information, such as transcribing a recorded session, we say so in our consent and privacy materials. When participants interact directly with a chatbot, AI moderator, AI prompt, or another AI system, we clearly tell them that they are interacting with AI rather than a person. Participants may ask us how their information is captured, handled, and stored, and may withdraw at any time.

  • Lawful data handling: Data and inputs used by AI systems must be lawfully collected, used, and disclosed, with applicable privacy rights and participant expectations respected throughout the research.

  • Protection from harm: We use AI in ways that protect participants from harm, including embarrassment, discrimination, or loss of privacy. Responses are reported in aggregate or in a de-identified form unless a participant has explicitly agreed otherwise.

  • Children and youth: We do not use AI tools to interact directly with participants under 16, and AI-assisted processing of their information is limited to what is necessary for the study and is subject to the same consent requirements described in our Privacy and Confidentiality Statement.

  • Recordings: Recordings and transcripts remain subject to our existing consent, access, and retention rules, and are only shared with clients with explicit informed consent from participants.

6. Protecting client information

  • Information gathered or developed on behalf of a client is used only for the services described in our agreement with them.

  • Client names, proprietary ideas, and confidential materials are removed or masked before any content is entered into an AI tool, wherever the task allows.

  • Access to AI-assisted work is limited to the people who need it for the project.

  • We will discuss any intended AI use with the client and obtain their agreement before using AI on their project, and will document it in the project agreement or proposal.

7. Data security and approved tools

  • We only use AI tools that have been reviewed and approved by Fraser Insights, and that meet our standards for data privacy and security. Before approving a tool, we consider whether it:

  • Does not use our inputs or outputs to train its models, and does not retain them longer than needed to deliver the service;

  • Provides encryption in transit and at rest, access controls, and multi-factor authentication;

  • Offers clear contractual confidentiality and security commitments, and is transparent about where data is processed and stored;

  • Is subject to due diligence before use and regular review of the provider's security policies, safeguards, and compliance status;

  • Supports ongoing security monitoring and periodic audits that identify risks arising from AI use;

  • Is covered by incident response procedures that specifically address AI-related security breaches; and

  • Can be configured to meet any additional requirements set by a client.

  • Where legal requirements, CRIC standards, client contracts, or Fraser Insights requirements differ, we follow the requirement that sets the highest bar for data security. Free or consumer versions of AI tools that do not meet these standards are not used with client or participant information. Approved tools are reviewed periodically, and removed from use if they no longer meet our standards. Our broader security practices, including device security, secure file transfer, and Canadian data storage in our Microsoft 365 environment, are described in our Privacy and Confidentiality Statement.

8. Reducing bias and ensuring quality

AI tools can reflect biases in the data they were built on, and can produce errors or content that sounds convincing but is wrong. To manage this risk:

  • Researchers review AI-generated questions, codes, themes, and summaries, and check them against the source data before they are relied upon.

  • We look for wording, assumptions, or patterns that could disadvantage or misrepresent individuals, communities, or equity-deserving groups, and correct them.

  • We periodically audit material AI uses for bias, inaccuracies, and uneven impacts, and use controlled test environments where appropriate before relying on a new tool, model, or workflow in live research.

  • Where AI-assisted analysis informs a finding, a researcher can trace and explain how the finding was reached.

  • Content produced predominantly or entirely by AI is explicitly identified as such.

  • We do not rely on AI alone for sensitive, high-stakes, or nuanced interpretation.

9. Transparency with clients

We are open about how we use AI. We distinguish insights produced solely by AI from insights generated through human analysis, whether or not that analysis was assisted by AI. For each project, we will explain, as applicable, the role of AI in fieldwork, data cleaning, analysis, and report generation; clearly identify whether synthetic personas or synthetic data were used and to what extent; and describe how human oversight was applied to verify or refine AI-generated content and insights. On request, we will tell clients which types of AI tools are used on their project, what tasks they support, and what safeguards apply. If our use of AI on a project changes, we will let the client know before the change takes effect.

10. Retention and destruction

Information processed with AI tools is kept only as long as needed for the research purpose and any contractual or legal requirement, in line with the schedule in our Privacy and Confidentiality Statement. Where a tool stores inputs or outputs, we delete them when the work is complete, or ensure the tool's settings do not retain them.

11. If something goes wrong

If an AI tool is involved in a security incident or a suspected misuse of information, we will act quickly to contain it, assess the risk, and keep a record of what happened. If there is a real risk of significant harm, we will notify affected individuals, the Office of the Privacy Commissioner of Canada, and affected clients, as required.

12. Questions and concerns

Clients and participants can contact us with questions about how we use AI, or to raise a concern. We will respond within ten business days. Contact: Erik Fraser, erik@fraserinsights.ca

13. Changes to this policy

AI technology and standards are changing quickly. We review this policy at least annually, and sooner if our practices, the tools we use, or legal and industry requirements change. Last updated: October 2026.

Reference: CRIC Guiding Principles for AI use in Market Research (July 2026).