AI Boom Built on Exploitation and Environmental Damage, UN Researchers Find

AI Boom Built on Exploitation and Environmental Damage, UN Researchers Find

User avatar placeholder
Written by Nan Hubbard

May 1, 2026

The rapid expansion of artificial intelligence is built on a supply chain that relies heavily on environmental degradation and labor exploitation in some of the world’s poorest regions, according to researchers and UN agencies.

Global map showing water stress levels
Source: United Nations University Institute for Water, Environment and Health

In arid regions such as Chile’s Salar de Atacama, mining activities account for up to 65% of total regional water use, competing with agriculture and local ecosystems. Groundwater levels have dropped, salt lagoons have shrunk, and freshwater aquifers face increasing risk of depletion and contamination. Lithium extraction—the key input for EV batteries and AI data center power infrastructure—has displaced Indigenous communities, destroyed habitats, and exposed workers to severe health risks including respiratory problems and skin conditions. Flamingo populations in the region have declined by up to 50% as breeding grounds have been disrupted.

Family working at artisanal cobalt and copper mine in DRC
A family works at an artisanal cobalt and copper mine site in 2025 in Kolwezi, Democratic Republic of the Congo. These mines are often unregulated. Michel Lunanga/Getty Images

The AI industry’s reliance on data labeling workers represents another form of exploitation. Companies including Google, Amazon, and Microsoft depend on human annotators primarily in the Philippines, Kenya, and Ukraine to categorize and label the data that trains AI models. These workers typically earn between $1 and $3 per hour. Some report psychological trauma from processing disturbing content, including images of murder, suicide, and sexual abuse.

The Cost of Clean Energy Transition

A 2024 UN Environment Programme report emphasized that the transition to clean energy must not come at the expense of human rights. The report calls for stronger regulations and corporate accountability to ensure that benefits of the green economy are shared equitably and that affected communities have a central voice in decision-making.

AI infrastructure—particularly the data centers required to train and run large models—consumes enormous amounts of electricity and water. The mining of rare earth minerals and metals needed for servers, networking equipment, and cooling systems creates environmental and social costs that are largely invisible to consumers of AI services.

Chile tailings pond from copper mining
Copper-mining companies create huge tailings ponds to store toxic byproducts of mining. Hundreds of these waste ponds exist across Chile and carry the risk of leaking acidic water and heavy metals such as arsenic, copper and mercury into groundwater. Martin Bernetti/AFP via Getty Images

Water pollution compounds these challenges. Mining generates large quantities of toxic byproducts, including chemical compounds that contaminate waterways and soil, affecting communities downstream from extraction sites.