AI and Indigenous Data Sovereignty: What OCAP and CARE Say

What OCAP and the CARE Principles ask of AI projects that use Indigenous data, and what Canadian research rules and privacy law do and do not cover.

Guide: AI BasicsLast verified: October 20266 min read

Indigenous data sovereignty is the right of Indigenous Peoples to govern data about their communities, lands, cultures, and knowledge, and it reaches AI because models are built from data. Two frameworks give the idea practical content: the First Nations principles of OCAP (ownership, control, access, and possession), which date to 1998, and the CARE Principles for Indigenous Data Governance, published in 2020. Neither is a statute, but Canadian research ethics rules refer to OCAP, and both frameworks pose direct questions to anyone who wants to train an AI system on data that describes Indigenous people.

Key takeaways

  • OCAP was developed by First Nations in Canada and is now maintained by FNIGC, the national First Nations information governance organization.
  • The CARE Principles (collective benefit, authority to control, responsibility, ethics) were written to sit beside the FAIR data principles, which say little about Indigenous rights.
  • The Tri-Council Policy Statement (TCPS 2, 2022) requires review and, in some cases, community engagement before Indigenous data is reused for research.
  • Canadian private-sector privacy law was not written around collective data rights, and an FNIGC project funded by the federal privacy commissioner's office critiques that gap.

Two frameworks with different origins

The Canada School of Public Service describes OCAP as a framework that First Nations developed through a National Steering Committee in 1998 and that FNIGC maintains today. Ownership means a community owns its information collectively. Control is the authority to decide how data is handled. Access covers who can see and manage the data. Possession is the physical control of data, which lets ownership be asserted in practice. The framework is specific to First Nations, and each community can interpret it for itself.

The CARE Principles have a different origin. A group of researchers working with the Global Indigenous Data Alliance published them in the Data Science Journal on November 4, 2020 (Carroll and 13 co-authors). They were designed to complement the FAIR principles, which ask that data be findable, accessible, interoperable, and reusable. FAIR focuses on making data easy to share. CARE adds a question about who benefits and who decides.

OCAP and CARE side by side

FeatureOCAPCARE
ScopeFirst Nations in CanadaIndigenous Peoples internationally
Four partsOwnership, control, access, possessionCollective benefit, authority to control, responsibility, ethics
Main concernWho holds and decides about community dataHow data ecosystems should serve Indigenous Peoples across the data lifecycle
Relation to FAIRNot framed as a companion to FAIRDesigned to be used together with FAIR
Legal statusA standard maintained by FNIGC, referred to in TCPS 2A set of principles in a peer-reviewed article

The table is a simplified summary. The primary documents define each term in more detail.

Where AI enters the picture

The CARE article itself names the tension. Its abstract points to concerns about secondary use of data and about how Indigenous data rights sit with open data, machine learning, broad data sharing, and big data initiatives. A model trained on scraped text, images, or recordings can carry Indigenous knowledge or personal information far from the community that produced it, and the original contributors usually have no say in that reuse. Under CARE, the ethics principle asks that Indigenous people be able to assess benefits, harms, and potential future uses of their data. The responsibility principle asks users to build respectful relationships with the source communities.

The Canada School of Public Service article published on January 8, 2025 makes the same link from a government angle: AI systems depend on data, and data about Indigenous Peoples should be governed in ways that respect their rights. The article also notes that federal information laws do not always reflect Indigenous approaches to collective data governance.

What Canadian research rules say about reuse

Chapter 9 of the Tri-Council Policy Statement (TCPS 2, 2022) governs research involving First Nations, Inuit, and Métis Peoples. It applies to research funded by the federal research councils, so it is not a general rule for companies. Article 9.8 asks researchers to learn about community customs and codes of research practice, and the accompanying text notes that many First Nations have adopted OCAP. Article 9.20 requires ethics board review for secondary use of identifiable Indigenous data, and community engagement before use when the use is not covered by an agreement or the original consent and the data is not publicly available. Under article 9.21, publicly available information generally does not require community engagement, though culturally informed advice is encouraged.

That last point matters for AI. Data that is technically public, such as posts or published recordings, may still fall outside what a community would accept, and the policy leaves that judgment to engagement and advice rather than a fixed rule.

The privacy law gap

An FNIGC project funded through the Office of the Privacy Commissioner of Canada's Contributions Program in 2022 produced an issue paper that analyzes the Personal Information Protection and Electronic Documents Act (PIPEDA) through the lens of First Nations data sovereignty and OCAP. According to the commissioner's summary, the paper ends with a critique of Canadian private-sector privacy law and a possible roadmap for reform. The summary does not list detailed findings, and it does not mention AI, so the full paper is the place to look for specifics.

Questions to ask before using Indigenous data in an AI project

  • Which community produced the data, and has that community been asked?
  • Is there a research agreement or data-sharing agreement that covers machine learning, not only the original study?
  • Who holds the data, and can the community withdraw it later?
  • Would the model's outputs reveal or reproduce sensitive cultural knowledge?
  • What benefit returns to the community, and who decides how that benefit is measured?
  • Does the data come from a source that is public but was never intended for reuse?

This article provides general information and is not legal advice. It covers Canadian and international frameworks. Tribal data governance in the United States was not verified here, and project teams with legal questions should consult qualified counsel and the communities concerned.

FAQ

Is OCAP a law?

No. OCAP is a standard that First Nations developed and FNIGC maintains. It gains legal weight when governments, funders, or agreements incorporate it, as TCPS 2 does by referring to it.

Does CARE replace the FAIR principles?

No. The authors describe CARE as a complement to FAIR and aim for data stewards and users to apply both together.

Can an AI developer use any data that is publicly available?

Public availability does not settle the ethical question. TCPS 2 treats public information more lightly for research purposes, but it still recommends culturally informed advice, and CARE asks about benefits and harms regardless of availability.

Do these frameworks apply outside Canada?

CARE was written for international use. OCAP is specific to First Nations in Canada, although other Indigenous groups have developed their own data governance approaches.

Last verified: October 2026

Sources

  • Indigenous Data Sovereignty (DDN3-A11), Canada School of Public Service, January 8, 2025 (updated October 19, 2026)
  • Carroll SR et al., The CARE Principles for Indigenous Data Governance, Data Science Journal 19(1):43, November 4, 2020
  • Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans, Chapter 9, 2022
  • First Nations Data Sovereignty and PIPEDA, Contributions Program project summary, Office of the Privacy Commissioner of Canada, 2022

New to artificial intelligence?

Start with the basics, then follow the guides to tools, rules, and effects on society.

Start with AI Basics