Emerging sovereign systems: global examples of indigenous knowledge machine language models [Forum special series, Artificial intelligence]

This article is part of a special series of summaries of keynotes and presentations from the KM Triversary Forum 2025, and also part of the artificial intelligence series. 

This KM4Dev Triversary Forum presentation, “Local AI in Action: Global Examples of Indigenous Knowledge Language Models,” systematically builds a case for a structural shift in AI development by detailing the theoretical foundations, the supporting policy arc, with example bridges to current on-the-ground practices.

The presentation briefly describes the policy arc emphasizing the evolution (2021-2025) toward indigenous knowledge sovereignty (IDS) and the explores challenges associated with low-resource languages (LRL). The ethical and technical requirements for localized AI are anchored in IDS, which mandates the principles of consent, credit, compensation, and control (the 4 Cs).

Five inclusion pillars: The proposed framework for this structural shift is organized around five critical pillars: pluralism, epistemic inclusion, co-governance, institutionalization, and ethics.

Emerging global examples (reality bridges): The work highlights operational examples of indigenous-led AI development, demonstrating tangible shifts from extractive models to sovereign systems:

• Aing A.I. (Inuktitut, Canada):  This project is a model for IDS, developing an interpretation and translation app to make Inuktitut accessible while ensuring the community retains control and copyright over linguistic data.
• Egune AI (Mongolia) and Lelapa AI (South Africa):  These initiatives focus on addressing culture and authentic data, with Egune AI emphasizing the creation of a national AI infrastructure and the vital use of synthetic data for LRLs.
• Sahabat-AI (Bahasa Indonesia):  This open-source ecosystem promotes digital sovereignty by empowering cultural nuances within a nationally specific context.

The role of ontologies: A crucial monitoring development involves the creation of ontologies for trustworthy indigenous knowledge AI (IK-AI). These are presented as a cultural and ethical necessity, providing a structured framework to formally represent Indigenous worldviews and counteract systemic bias inherent in Western-centric datasets.

A structural shift:  Achieving truly beneficial AI requires a fundamental transition from indigenous knowledge systems as passive subjects of data extraction to active, sovereign agents who co-create and govern AI systems.

Biography:

Elizabeth (Liz) McLean is an experienced knowledge and information manager with a current focus on humans first KM+AI roles and roadmaps. Knowsaic, her Washington DC region consultancy, has provided KM, IM, and IA services to federal clients, international development agencies, implementing partners, and arts organizations. Liz currently serves as a KM Practitioner Mentor with the KM Peer Mentoring Program (KMPM). Learning over the last 18 months includes knowledge engineering, ontologies, and AI’s implications for organizational strategies. She is currently digging into the essential roles that KM professionals need to play in shaping human in the loop (HITL) + AI using traditional KM approaches. She evaluates and assesses LLMs for usability and quality of information architecture. APQC recently featured her work on the role of concept mapping as a first step on the AI enabling roadmap, which demonstrates that KMers who may not be AI experts, are able to shape and embed an organization’s learning and innovating path toward AI readiness.

Presentation resources: PowerPoint slides.

Header image source: GoTo Press Release announcing the launch of Sahabat-AI.

AI statement: Google’s NotebookLM was used to store sources found by the author to query sources for concepts and co-locate collection resources.

References and further reading:

1. United Nations (UN) policies: 

UNESCO. (2001). Universal Declaration on Cultural Diversity. Paris: UNESCO.

UNESCO. (2003). Convention for the Safeguarding of the Intangible Cultural Heritage. Paris: UNESCO.

UNESCO. (2005). Convention on the Protection and Promotion of the Diversity of Cultural Expressions. Paris: UNESCO.

UNESCO. (2021). Recommendation on the Ethics of Artificial Intelligence. Paris: UNESCO. 

United Nations. (2007). United Nations Declaration on the Rights of Indigenous Peoples (UNDRIP). New York: United Nations.

2. Other references and resources:

CIA. (2025), Maps—The World Factbook. https://www.cia.gov/the-world-factbook/maps/ (Note: the CIA World Factbook has been discontinued, but past editions up to February 2026 are still accessible in online archives, https://onlinebooks.library.upenn.edu/webbin/serial?id=worldfactbook).

GoTo. (2024, November 15). Indosat Ooredoo Hutchison and GoTo Launch Sahabat-AI: Indonesia’s Open-Source LLM for Empowering Digital Sovereignty.  GoTo. https://www.gotocompany.com/en/news/press/indosat-ooredoo-hutchison-and-goto-launch-sahabat-ai-indonesias-open-source-llm-for-empowering-digital-sovereignty.

RELX SDG Resource Centre. (n.d.). Indigenous Knowledge. https://sdgresources.relx.com/indigenous-knowledge.

Sharma, R. (2025). Mongolian AI startup secures $3.5 million to drive national AI infrastructure. The Fast Mode. https://www.thefastmode.com/investments-and-expansions/41648-mongolian-ai-startup-secures-3-5-million-to-drive-national-ai-infrastructure.

Srreyansh, S. (2024). Advances and challenges in developing large language models for low-resource languages. Acta Scientific Computer Sciences, 6(5), 03–39. https://actascientific.com/ASCS/pdf/ASCS-06-0519.pdf.

Wat, S. (2025, August 28). Inuit innovators turn to AI to revitalize Inuktitut. CBC News. https://www.cbc.ca/news/canada/north/inuit-innovators-ai-inuktitut-1.7618536.

Zhou, V. (2025, August 22). The Mongolian startup defying Big Tech with its own LLM.  Rest of World. https://restofworld.org/2025/mongolia-egune-ai-llm/.