AI is a double-edged sword for Indigenous land protection, UN experts warn

This story is published through the Indigenous News Alliance.

Artificial intelligence (AI) is increasingly being used by Indigenous communities to detect illegal logging, track wildfires, and monitor traditional lands. However, the data centers that power AI are creating new challenges, as they require substantial amounts of water, energy, and critical minerals, often sourced from Indigenous territories.

Currently, Indigenous leaders at the United Nations Permanent Forum on Indigenous Issues (UNPFII) are grappling with a dilemma: how to leverage the protective potential of AI while avoiding the extractive practices they have opposed for generations.

A recent study by Hindou Oumarou Ibrahim, a Mbororo and former chair of the permanent forum, explores the opportunities and challenges AI presents for environmental protection, as well as its impacts on Indigenous territories. Issues such as land-grabbing, water overexploitation, and land degradation are highlighted due to the technology’s significant energy, water, and mineral demands.

“For generations, Indigenous Peoples have protected the world’s most intact ecosystems without satellites, without algorithms or technologies,” Ibrahim told Mongabay. “AI can become a powerful ally to that stewardship, if it is used on our terms in a culturally appropriate way.”

Ibrahim pointed out that AI can assist Indigenous communities in monitoring biodiversity, detecting deforestation, illegal mining, wildfires, or water contamination using satellite imagery and sensors. “When combined with Indigenous Peoples’ knowledge, AI can help predict climate impacts, track wildlife movements, and strengthen land-use planning while helping to plan faster resilience strategies,” she added.

In Brazil’s Acre state, the Katukina/Kaxinawá Indigenous Reserve’s agroforestry agents have been employing AI to combat deforestation. The reserve is listed among the top five for deforestation risk, as per a forecast by an AI tool developed by Microsoft and the Brazilian nonprofit Imazon.

“It is very important to monitor the land because we Indigenous people are safer when we can detect if someone is invading, if someone is taking wood from our land, if someone is hunting directly on our land, if someone is putting up a fire close to our land,” Siã Shanenawa, one of the 21 agroforestry agents in the reserve, explained.

Lars Ailo Bongo, a professor at UiT The Arctic University of Norway, leads the Sámi AI Lab, which explores how AI can assist Indigenous Sámi people. Despite AI not being inclusive enough, he sees opportunities for communities. “AI can democratize access to the analytical capabilities needed to conduct data-driven modeling aligned to Sámi views and norms,” he noted.

In Nunavut, Inuit communities are merging traditional knowledge with predictive AI models and time-series analyses to identify new fishing locations amid climate change impacts on fish availability. Similarly, in Chad, Indigenous pastoralists combine participatory mapping and satellite data with predictive AI tools to foresee severe droughts and secure transhumance corridors, enhancing their climate resilience.

In South America, Rainforest Foundation US utilizes a blend of traditional knowledge and emerging technologies, from planting trees along boundary limits to using smartphones and drones, to aid Indigenous communities in safeguarding their territories.

“AI is the latest tool in that continuum,” Cameron Ellis, field science director at Rainforest Foundation US, stated. “Community monitors can use AI-derived remote sensing products to analyze large volumes of satellite data and interpret deforestation patterns linked to mining or agriculture expansion, responding more swiftly to these threats.”

Residents and farmers in Thailand’s Chonburi and neighboring Rayong province, experiencing water shortages and pollution, have expressed concerns about the environmental impacts of data center expansion—integral to AI infrastructure. Data centers consume large amounts of water for cooling and significant energy to operate.

This scenario is mirrored in several communities globally, from rural areas in eastern Pennsylvania to villages in Querétaro, Mexico. Residents are concerned about wastewater contamination, water and energy shortages, and rising costs associated with data center expansion.

“AI is often perceived as immaterial, but it carries a very real environmental footprint,” Ibrahim pointed out. “It depends on vast amounts of energy, water, and critical minerals, many of which are extracted from or located near Indigenous Peoples’ territories, leading to land degradation, biodiversity loss, and, in some cases, community displacement.”

Beyond the environmental impact of data centers, Ibrahim’s study also highlighted other challenges Indigenous peoples face regarding AI, such as insufficient infrastructure, legal protection, and institutional capacity to safeguard digital rights. She noted that AI could lead to the exclusion of Indigenous peoples or facilitate the extraction of sensitive data. The use of drones, satellites, or mapping tools without prior consultation with Indigenous peoples could expose the location of sacred sites, ecologically strategic areas, or other sensitive locations.

Kate Finn, a citizen of the Osage Nation and executive director of the Tallgrass Institute, which works to align investor strategies with Indigenous rights, speaks of an “opportunity space” within AI to help Indigenous peoples preserve their languages and enhance their governance systems. She also shares concerns about the environmental risks. “The consistent ask from Indigenous peoples around the world is that they want their free, prior, and informed consent respected before data centers go into their land,” she emphasized. “As we approach AI from an Indigenous lens, it will necessarily have to take account of all of those different nodes, both the opportunity space, but also a protective space of lands, territories, and resources, and also of language and culture, and the creative property that Indigenous peoples have placed online.”

Bongo stated that the Sámi face limitations due to a lack of funding to hire AI developers who can create Sámi-aligned AI models and make these available to the community. “This is especially sad, since we have Sámi AI developers that are interested in doing the work,” he explained, highlighting a lack of capacity, not competency. “To make progress, there is a need for a bigger center and push, which the Sámi organizations do not have the budgets for, so the states need to provide the funding [Norway, Finland, and Sweden].”

For projects that rely on external funding, Bongo emphasized the importance of ensuring that Indigenous peoples do not become a small minority partner.

“Technology on its own doesn’t protect forests – people do,” Ellis remarked. “These tools are only effective when grounded in community governance and leadership, and when the data they generate is used to trigger action on the ground. Likewise, communities must be able to retain sovereignty over how their data is collected to ensure it advances their own priorities without undermining their rights.”

Ibrahim stressed that to protect Indigenous peoples and their territories, governments must prevent all forms of land-grabbing, water exploitation, and mining activities related to data centers and energy sources, while respecting Indigenous rights, worldviews, and aspirations.

“AI becomes harmful when it is imposed without free, prior, and informed consent,” Ibrahim stated. “In that context, it risks repeating old patterns of extraction of the resource, data, and appropriation of knowledge and the credit to these knowledge.”