Photo: A red sand dune with spinifex and saltbush under a wide blue sky. Simpson Desert, Australia. Neiste, CC BY-SA 4.0, via Wikimedia Commons
Nancy E. Weiss argues that voice, agency and authority offer a way to think about Indigenous knowledge in the age of AI, and that the real question is on whose terms knowledge will flow.
AI didn’t create the challenge of caring for Indigenous knowledge, particularly knowledge collected and documented outside the communities from which it originates. But it changes the ease with which that material can be gathered, processed, recombined, and disseminated at scale. It also makes more pressing the question of what responsibilities and obligations accompany the use of Indigenous knowledge.
Dr Terri Janke, the Australian Indigenous lawyer and scholar, has expressed the challenge posed by AI rather memorably:
AI has got no Dreaming. It is not responsible for having any kinship, country or cultural obligations.
A collection may be publicly accessible. A particular use may be permitted under copyright law. And technology may make it extraordinarily easy to ingest, reproduce, transform, and disseminate the material. But neither law nor technology necessarily answers the question of what responsibilities and obligations are due to the communities from which that knowledge comes.
In the United States, as in many other countries, we operate simultaneously within multiple legal frameworks: Tribal, state, national, and international. For many reasons, it is neither possible nor prudent to isolate Indigenous knowledge completely. We live together and learn from one another. So how do we engage responsibly in this moment of technological development?
In describing the importance of recovering and preserving endangered knowledge, Dr Martin Nakata observed that knowledge recovery led by Indigenous communities may look very different from knowledge recovery led by scientists, technologists, conservationists, and others. In this regard, it is helpful to see how Indigenous communities, scientists, technologists, and conservationists came together to identify guidance for the use of traditional knowledge in government policy-making: Guidance for Federal Departments and Agencies on Indigenous Knowledge.
Voice, agency and authority
I find it useful to think about three frameworks of engagement: voice, agency, and authority.
By Voice, I mean Indigenous peoples defining and describing their own knowledge, rather than others speaking for them. It is about authenticity, representation, and whose perspectives are heard.
Agency is the ability to participate meaningfully in decisions about how knowledge flows: how it is collected, described, digitised, used, and reused. In the United States, the Native American Graves Protection and Repatriation Act (NAGPRA), for example, provides for the return of Native American remains, certain cultural items, and objects of cultural patrimony, establishing legal requirements for consultation, a way of promoting agency. Federal agencies also have consultation responsibilities with Tribes and other Indigenous communities on policies affecting them.
Authority asks who has control; whose decisions govern? Who has the power to establish the terms?
Tribal laws, international frameworks, national laws, institutional governance, and platform rules all come into play. These are distinct sources of authority, with different reach and effect. A platform rule is not a Tribal law; an institution’s policy is not an exception under copyright. We need to understand what each governs, how they relate to each other, and what happens when their answers diverge.
AI systems depend on data. Questions about Indigenous knowledge and AI therefore inevitably become questions about the governance of the data and knowledge on which those systems depend. AI governance requires data governance.
Indigenous data sovereignty asserts the rights of Indigenous peoples to govern data concerning them, grounded in responsibilities to their peoples, knowledge, lands, and resources. The Global Indigenous Data Alliance and the U.S. Indigenous Data Sovereignty Network have developed practical tools and resources in furtherance of data sovereignty and its implementation through data governance.
And there is much happening on the ground, so to speak. Dr Evelyn Cox has chronicled the ways in which Tribal Nations are putting these ideas into practice through Indigenous-controlled data infrastructure, archival data sovereignty, and Tribal AI policies and governance systems.
The emerging work I hear about from Indigenous librarians, archivists, museum professionals, artists, technologists, and knowledge keepers is encouraging. It is important not only for protecting Indigenous knowledge, but also for ensuring authentic voices are heard in media, literature, scholarship, and other forms of knowledge production.
Tools we already have
We need to look at cultural heritage law and intellectual property (IP) systems together. Intellectual property can do some things, even if it cannot address every concern. At American University’s Washington College of Law, students in the Glushko-Samuelson Intellectual Property Law Clinic work on a variety of matters to help traditional knowledge holders retain control over, and secure legal rights in, their work products.
My former employing agencies, the U.S. Institute of Museum and Library Services and the National Endowment for the Humanities, provided early support for Mukurtu, a content management system through which Tribes manage and share cultural materials according to community-defined protocols. Mukurtu creates space for differentiated access, cultural context, community knowledge, and relationships around materials. It demonstrates how a digital system can accommodate culturally relevant protocols for managing access to Indigenous knowledge.
An archive and a trained AI model work very differently, however, so we cannot assume these controls transfer directly. But Mukurtu offers a useful starting point for thinking about what AI systems would need to recognise and accommodate culturally relevant distinctions.
Legal responsibilities
So yes, as Dr Janke has expressed, AI has no Dreaming, no kinship, no country, and no cultural obligations. But people and institutions do. The people who collect and preserve knowledge do. The institutions that digitise and provide access do. And the companies that build AI systems, assemble training data, and deploy those systems make choices about which obligations their technologies will, or will not, recognise.
When legal permission, technical capability, and community authority produce different answers, how should we decide what responsible AI use requires?
I don’t believe the answer is simply more access or less. Nor do I believe it is either realistic, or necessarily desirable, to place Indigenous knowledge entirely inside or outside the AI ecosystem.
The question we should continually ask is: On whose terms?
And perhaps “on whose terms” is ultimately another way of asking whether Indigenous peoples have not only voice, and not only agency, but also meaningful authority in shaping the answers.
This is an abridged version of the presentation by Nancy E. Weiss (American University Washington College of Law) at USERrights 2026, the Centre on Knowledge Governance conference held in Geneva from 30 September to 2 October 2026, in the session “Traditional Knowledge: Legal Rights and Ethical Responsibilities”.
To read and watch the USERrights 2026 presentations online, visit: https://knowledgegov.org/userrights2026/report/
