Introduction

Artificial intelligence (AI) tools—such as automated transcription, qualitative coding assistance, and text summarization—are increasingly used in human subjects research. These tools can streamline research workflows, particularly for studies involving interviews, focus groups, or large qualitative datasets. However, when research involves Indigenous populations or Indigenous data, the use of AI introduces additional ethical, legal, and cultural considerations. This blog post outlines important considerations for researchers planning to use AI tools in studies involving Indigenous populations, including how these technologies intersect with governance structures, participant protections, and community expectations regarding data stewardship.

Indigenous Data Sovereignty

A key starting point is recognizing that many Indigenous groups—such as federally recognized tribes in the United States—are sovereign nations with legal and political authority over research involving their citizens and data. Tribal sovereignty may include governance over how data are collected, stored, analyzed, and shared. In addition to institutional IRB requirements, some Indigenous nations maintain their own research review boards or approval processes. Frameworks such as the CARE Principles for Indigenous Data Governance emphasize that data concerning Indigenous peoples should be used in ways that support collective benefit and respect community authority over data stewardship. When AI tools are incorporated into research workflows, investigators should consider whether the platform stores uploaded data, retains transcripts, or uses submitted materials to train machine learning models, and whether community approval is required for secondary uses of data.

Community Consultation Before Using AI

Community consultation is another important aspect of ethical research practice. Some Indigenous nations maintain tribal IRBs or formal research review boards that review research involving their citizens or communities. When such structures exist, researchers should follow those procedures and obtain the appropriate approvals. However, not all Indigenous communities have formal research review boards. In such situations, researchers should still identify appropriate avenues for consultation before implementing research methods involving AI tools. This may include engaging with tribal leadership, community elders, tribal health or education departments, or community-based organizations that represent the population involved. Early consultation can help ensure that research practices align with community expectations regarding data stewardship, interpretation, and cultural sensitivity.

AI Data Sharing and Confidentiality Risks

Researchers should also evaluate confidentiality risks associated with AI tools. Platforms used for transcription, coding, or summarization may process data through external vendors or cloud-based systems. Before using these tools, investigators should understand how the software handles research data and whether it stores or retains uploaded materials. Particular caution is warranted when working with culturally sensitive materials such as oral histories, traditional knowledge, or community narratives. Researchers should avoid uploading identifiable or culturally sensitive information into public AI platforms unless appropriate safeguards and approvals are in place. When possible, transcripts and datasets should be de-identified prior to AI processing, and institutionally approved systems should be used whenever available.

See TC IRB's guide on Using Artificial Intelligence (AI) in Human Subjects Research. We’ve developed a comprehensive guidance document to support researchers in navigating AI use in human subjects research: TC IRB Guidelines for Ethical and Secure Use of AI-Based Software in Research

This document includes:

• Step-by-step recommendations for consulting with TCIT
• Requirements for disclosing AI use in IRB applications
• Considerations for free vs. paid tools
• A researcher's checklist and FAQs

Informed Consent and Transparency

Transparency about AI use is also an important component of informed consent. If automated tools will be used to process recordings or qualitative data—for example, through automated transcription or AI-assisted coding—participants should be informed during the consent process. Consent materials should describe what tools will be used, whether third-party vendors will process the data, and what steps are taken to protect confidentiality. Researchers should consult the TC IRB Informed Consent Form Template when developing consent language. For example:

Audio recordings may be transcribed using automated transcription software to produce text transcripts. Identifying information will be removed before analysis.

Researchers should verify the data storage and retention practices of the specific software they plan to use before making representations about how recordings or transcripts are handled.

Prevent Algorithmic Bias and Cultural Misinterpretation

Researchers should also consider the potential for algorithmic bias and cultural misinterpretation when using AI tools. Many AI models are trained on datasets that underrepresent Indigenous languages, histories, and lived experiences. As a result, automated transcription, coding, or summarization tools may misinterpret culturally specific language or narratives. AI-generated outputs should therefore be treated as assistive tools rather than definitive interpretations, and researchers should review and interpret results within appropriate cultural and research contexts.

Limit Secondary Use of Data

Another important consideration is limiting unintended secondary uses of data. Some AI platforms retain uploaded materials or use them to improve or train their systems. Investigators should review whether the platform uses submitted data for model training, whether transcripts or recordings are retained after processing, and whether future analyses would require additional approvals. When Indigenous data governance structures apply, additional community review or approval may be required before research data can be reused for new purposes.

Data Storage and Access Controls

Researchers should also ensure that appropriate data storage and access controls are in place when incorporating AI tools into research workflows. Research data should be stored on secure, institutionally approved systems and access should be limited to authorized research personnel. Investigators should also document how AI tools are used in the research process so that data handling practices remain transparent and consistent with institutional policies and any applicable community governance requirements.

Cultural Competency Training for Researchers

Cultural competency training can help researchers better understand these considerations. Programs such as rETHICS (Research Ethics Training for Health in Indigenous Communities) provide guidance on respectful engagement with Indigenous communities, the historical context of extractive research, and responsible stewardship of community data. Because rETHICS focuses primarily on health research contexts, researchers working in other fields—such as education, psychology, or the social sciences—may need additional or discipline-specific training to address the ethical considerations relevant to their research settings (see TC IRB Education & Training page).

Human Oversight of AI Analysis

Finally, AI-assisted analysis should always include human oversight. Automated tools can support transcription or coding processes, but they should not replace researcher judgment or community-informed interpretation of qualitative data. Investigators should review AI-generated outputs carefully, ensure that interpretations are grounded in appropriate cultural context, and consult community collaborators when relevant.

Conclusion

As AI tools become more common in research workflows, investigators should ensure their use aligns with both institutional policies and the governance structures and expectations of Indigenous peoples. Researchers planning to use AI tools in studies involving Indigenous populations are encouraged to consult with the IRB early in the protocol development process, verify the data practices of any AI software used, and complete appropriate cultural competency training before beginning data collection.