Skip to content
Mik Explore × Maah · Sacred Forest
FR EN PT
Note 1 · Indigenous × AI R&D · Mik Explore

Thematic research · June 18, 2026

Indigenous × AI R&D

Six-dimension cartography of the intersection between artificial intelligence and Indigenous peoples, prepared to feed Sacred Forest's tech strategy, data ethics and village second brain layer.

The essential

Three operational frameworks hold international consensus: CARE Principles, OCAP, and the Indigenous Protocol AI Position Paper. The underlying controversy — AI as revolution or colonizer? — is settled pragmatically by communities themselves: neither rejection nor naive adoption, but conditional Indigenization under FPIC, sovereign infrastructures, and locally fine-tuned small language models. Three concrete levers for Sacred Forest: Mukurtu CMS + Local Contexts TK Labels for governance, Mapeo + community drones for monitoring, self-hosted open-weight models to avoid Big Tech cloud lock-in. Key reference to mobilize: the Wíhaŋble S'a Center at Bard College (Suzanne Kite, $500k NEH grant 2024) is currently prototyping the sovereign data storage protocols.

All cited sources are clickable in the body of the document. Full bibliography at the bottom of the page: go to the 35+ references →

Section 1 / 6

Three international frameworks that hold consensus

Referenced by UNESCO, funding agencies, the academic world, and major NGOs whenever Indigenous data is at stake.

CARE

CARE Principles

GIDA · Research Data Alliance, 2019

Complement to the FAIR principles. Four pillars: Collective benefit, Authority to control, Responsibility, Ethics. Now the standard cited by academic institutions, NGOs, and funding agencies whenever Indigenous data is involved.

Official source → Full text (PDF, 2019) →
OCAP®

Ownership, Control, Access, Possession

FNIGC, Canada

Precursor to CARE, focused on First Nations in Canada. More prescriptive on the legal property dimension of data. Trademark of the First Nations Information Governance Centre.

Official source →
IP-AI

Indigenous Protocol AI Position Paper

Lewis et al. · CIFAR & Concordia, 2019

Outcome of 20 months of collective work, two workshops in Hawai'i. Five culturally situated vignettes (Anishinaabe, Coquille, Kanaka Maoli/Blackfoot, Lakota, Euskaldunak) plus a technical prototype in 'ōlelo Hawai'i. Foundational reference of the field.

Official source → EN version (Concordia PDF) →

Schema — The 4 CARE pillars

How CARE complements FAIR

C

Collective Benefit

Generated value benefits communities equitably

A

Authority to Control

Right to Free, Prior and Informed Consent (FPIC)

R

Responsibility

For positive relationships, expanded capabilities, respected languages and worldviews

E

Ethics

Minimize harm, justice, UNDRIP alignment

UN framework — UNDRIP applied to AI

5 directly actionable articles

  • Art. 18Right to participate in decisions through chosen representatives — AI governance
  • Art. 25Relationship to the land — compute footprint of datacenters
  • Art. 29Prohibition of hazardous material storage without consent — datacenters on ancestral lands
  • Art. 31Right to maintain and protect cultural heritage and traditional knowledge — corpora, languages, sacred imagery
  • Art. 32Right to determine strategies for lands and resources — data is a resource
Full UNDRIP text (UN) →

UN standard

FPIC — Free, Prior, Informed Consent

Not a one-off event but a continuous process: free (no coercion), prior (before collection/use), informed (in language and terms understandable to the community), consent (capacity to refuse).

« No use of Indigenous data or heritage should ever happen without Free, Prior and Informed Consent. » — Cultural Survival, 2025. Includes training data, deployment, monetization, derivatives.

FPIC Manual — UN/FAO 2016 → Cultural Survival 2025 →

Section 2 / 6

Three documented major risks

The pitfalls a programme like Sacred Forest must name and defuse from its public posture.

01

Datacenters and extractive colonialism

Datacenters target rural and tribal areas for cheap water, cheap electricity, and tax incentives. +267% energy price for local consumers in Virginia. 6 billion gallons of Google water in 2023.

Source — Honor the Earth — Data Centers Myth vs Fact →
02

Cultural misrepresentation — Maasai case, 2024

Documented by Cultural Survival: Kenya's State Department of Culture published AI-generated images of Maasai attire in 2024 where men wore a necklace traditionally restricted to women. « This was not merely an aesthetic oversight. It was a cultural transgression. » Direct violation of UNDRIP art. 31. The kind of error a community-validation protocol would have caught.

Source — Cultural Survival — Indigenous Peoples and AI →
03

Data colonialism and toxic translation

Abeba Birhane (Mozilla) theorizes the algorithmic colonisation of Africa. Direct application: for the translation of Amazonian languages, available corpora often come from colonial missionaries — classified as potentially toxic data by IBM Brazil / USP researchers, unusable without explicit community agreement. Critical precedent for any Pano NLP project (Huni Kuin / Kaxinawa).

Source — Birhane — Algorithmic Colonisation →

Section 3 / 6

Seven operational tools ready to mobilize

Nothing to invent from scratch — everything exists, some have been deployed for 15 years in the Amazon, Australia, Africa, the Pacific. Click any name for the tool's reference.

Tool Sacred Forest relevance
Mukurtu CMS ↗ Foundational brick for the « village second brain »
Local Contexts TK Labels ↗ Ready-to-plug labelling system
Mapeo (Awana Digital) ↗ Already deployed in the Amazon with the Seikopai (Ecuador)
Rainforest Connection Guardian 3 ↗ Drone monitoring complement
Mistral / DeepSeek / Phi-4 ↗ Avoids Big Tech cloud lock-in
OpenDroneMap (ODM) ↗ Pairs with consumer DJI for the drone pillar
InkubaLM-0.4B (Lelapa AI) ↗ Transposable architecture for Pano

Section 4 / 6

Four precedent cases — Amazon and Pacific

Sacred Forest is not stepping onto virgin territory. These comparable initiatives have already documented what works and what breaks.

Brazil — Rondônia, Acre

WWF Brazil — 25 drone kits delivered in 2022

Programme started in 2019. 5 Indigenous communities equipped. Confirmed beneficiaries: Uru-Eu-Wau-Wau (victims of land grabs and timber trafficking). Extends surveillance range, avoids direct contact with criminals — 19 land defenders killed in Brazil in 2021.

Sources: InsightCrime ↗ Rainforest Foundation ↗ Mongabay ↗
Lead to explore: Felipe Spina (WWF Brazil)

Peru — Loreto

36 communities equipped with drones

Centralized database for environmental crimes (Mongabay 2020). Operational regional model, directly transposable to the Acre terrain.

Sources: Mongabay 2020 ↗

Aotearoa (New Zealand)

Te Hiku Media — economic precedent

Refusal of a Lionbridge offer at $45/hour to transcribe hours of Māori speech. Position: only the Māori people shall benefit financially from its own language. They created their own licenses. Strong precedent for the Huni Kuin / Arhuaco facing any external AI solicitation.

Sources: Te Hiku Media ↗

Brazil — São Paulo

IBM Brazil + USP — Guarani Mbya and Nheengatu

Explicit co-creation, paused on Mbya as long as there is no community consensus. Direct Latin-American precedent for the model to build for Pano languages (Huni Kuin, Kaxinawa).

Sources: AmericasNLP 2025 (ACL) ↗

Section 5 / 6

Four actors to explore in priority

Exploration leads and first doors to open for phase 0 of the ACRE programme.

Dr. Suzanne Kite

Oglala Lakota · Bard College

wihanblesa@bard.edu ↗

Director of the Wíhaŋble S'a Center — the first US center labelled NEH and led by American Indians on AI ($500,000 NEH grant, 2024). Currently developing sovereign data storage protocols, modelled on the Māori work. Co-author of the foundational text Making Kin with the Machines (MIT 2018).

Site: kitekitekitekite.com ↗ Chronogram profile (March 2026) ↗

Living, articulate reference. Potential informal mentor. Potential character for the documentary.

Local Contexts Hub

Multi-tribal partnership

To create a sandbox account and test TK Labels on 3-5 pilot Sacred Forest assets. To set the labelling protocol from Mission 1 onwards.

Site: localcontextshub.org ↗ TK Labels catalog ↗

Awana Digital / Mapeo

International, formerly Digital Democracy

For an offline-first app demo and Amazon deployment resources. Direct precedent with the Seikopai in Ecuador, in partnership with Amazon Frontlines + Alianza Ceibo.

Site: awana.digital ↗ Seikopai (Ecuador) case ↗

Mukurtu Support

WSU + Warumungu

support@mukurtu.org ↗

For an Acre-style deployment overview, training resources, hosting options. Open-source platform for Indigenous cultural archive management, deployed by hundreds of communities worldwide.

Site: mukurtu.org ↗ Mellon Foundation grant story ↗

Section 6 / 6

Four readings to prioritize

Making Kin with the Machines

Lewis, Arista, Pechawis, Kite — MIT JoDS 2018

Founding text of the Indigenous AI field.

Open ↗

Indigenous Protocol AI Position Paper

Lewis et al. — CIFAR / Concordia 2019

Full English PDF version.

Open ↗

CARE Principles for Indigenous Data Governance

Carroll et al. — Data Science Journal CODATA 2020

The reference academic paper.

Open ↗

Indigenous Peoples and AI

Cultural Survival 2025

Defending Rights, Shaping the Future of Technology. UNDRIP × AI applied.

Open ↗

Full bibliography

Sources & references

The entire set of links consulted for this research, classified by category. All links open in a new tab.

Doctrine, frameworks and standards

Critical voices and Indigenous perspectives

Operational tools

Amazon cases — drones and Indigenous monitoring

NLP, Indigenous languages and small language models

Thematic scan from June 18, 2026 — 10 Firecrawl queries + 9 deep reads + 30+ secondary sources. The raw research note (38 markdown pages, 17 KB) is kept in the Mik Explore knowledge base and available on request.

Next

The companion note — Debrief & Vision Mik × Maah

What came out of our three June conversations and the operational vision we sketched together.

Read note 2 →