Artificial intelligence (AI) is often heralded as a force of progress, driving innovation, economic growth, and unprecedented efficiency. Tech giants boast of AI’s potential to revolutionize industries, boost productivity, and even tackle pressing global challenges like climate change. But beneath this utopian narrative lies a darker reality—one where the economic rewards of AI are concentrated in the Global North, while its labor exploitation and environmental destruction are outsourced to the Global South. From the exploited workers behind AI training datasets to the environmental costs of massive data centers, the expansion of AI is reinforcing historical patterns of inequality. Rather than creating a democratized technological future, AI is deepening the global divide—what I term AI colonialism—where the benefits accrue to a select few while the burdens are externalized to the most vulnerable.
Despite the perception that AI operates autonomously, the technology relies heavily on human labor—specifically, low-wage workers in the Global South who perform data labeling, content moderation, and other tedious digital tasks. In countries like Kenya, India, and the Philippines, millions of workers sift through vast amounts of data to train AI models, earning as little as $1.50 per hour under precarious gig-economy conditions. The nature of their work can be grueling. Kenyan content moderators employed by subcontractors for platforms like Facebook and TikTok spend hours reviewing violent and disturbing material, often suffering from psychological trauma with little to no mental health support. In India, AI trainers annotate images, transcribe text, and flag inappropriate content—all essential for refining machine learning algorithms—yet they are treated as disposable, denied stable contracts, fair wages, and legal protections.
While Silicon Valley executives reap enormous profits, the labor that fuels AI development remains invisibilized. AI is not simply a neutral technological tool—it is embedded within a global system of exploitation that mirrors past colonial labor structures, extracting value from the Global South while keeping its workers marginalized.
AI is not just built on cheap labor—it is also built on staggering environmental costs, disproportionately borne by developing countries. The training of large-scale AI models requires massive computational power, leading to high energy consumption and carbon emissions. A single AI model like OpenAI’s GPT-3 can emit as much CO2 as five cars over their entire lifetimes. This energy demand is driving the rapid expansion of data centers, particularly in regions where electricity and land are cheap—often in the Global South. Countries like South Africa, Indonesia, and Brazil have become hubs for AI infrastructure, but at a devastating cost. These data centers require vast amounts of water for cooling, exacerbating water scarcity issues, while their massive electricity consumption often depends on fossil fuels, increasing carbon footprints.
Meanwhile, the extraction of rare minerals for AI hardware—such as cobalt, nickel, and lithium—further entrenches environmental degradation. In the Democratic Republic of Congo, where over 70% of the world’s cobalt is mined, workers endure inhumane conditions in hazardous, unregulated mines, often with children among the labor force. Similar mining operations in the Philippines and Latin America have led to deforestation, water contamination, and forced displacements of Indigenous communities. These environmental consequences are not borne equally. The Global North benefits from AI’s conveniences and economic growth while the climate burden falls disproportionately on the Global South, whose communities already face severe climate vulnerabilities. This is the hallmark of necroexportation—a system where technological prosperity in one part of the world is sustained through the systematic harm of another.
Some efforts toward AI governance, however, are already underway. As the first comprehensive regulatory framework on artificial intelligence, the European Union’s AI Act aims to manage AI risks, ensure transparency, and regulate high-risk AI applications. Its jurisdiction, nonetheless, is limited to Europe, leaving out the vast majority of AI workers, resource suppliers, and communities affected by AI-driven environmental degradation. Similarly, voluntary AI principles from the OECD and UNESCO emphasize ethical AI but lack enforcement mechanisms, allowing major tech firms to continue their exploitative practices without consequence (OECD AI Principles).
Hence, a  truly fair AI system must move beyond regional regulation. Global governance efforts must seek to orchestrate all relevant stakeholders towards the following four objectives: First, enforce Global Labor Protections: AI should not be built on sweatshop-like working conditions in Kenya, India, or Venezuela. The International Labour Organization (ILO) must establish binding global AI labor standards, ensuring fair wages, occupational safety protections, and collective bargaining rights for AI workers.
Second, mandate Ethical Sourcing of AI Hardware: Cobalt, nickel, and lithium—critical components of AI infrastructure—must be ethically sourced, with strict human rights due diligence laws to prevent child labor, hazardous working conditions, and violent resource conflicts.
Third, regulate AI’s Carbon Footprint: AI’s environmental impact is worse than most industries admit. Data centers now consume more electricity than entire countries, and their emissions are 662% higher than reported by Big Tech. AI regulation must include carbon caps, mandatory transparency on emissions, and investment in carbon-neutral AI training techniques.
Fourth, ensure Technology Transfer to the Global South: The power over AI industry is concentrated in the hands of a few wealthy corporations in the Global North, while reinforcing technological dependence in developing countries. Instead of extracting resources and labor while keeping AI expertise confined to Silicon Valley, the Global South must be empowered through technology transfer agreements, AI research funding, and inclusive AI infrastructure development.
For too long, AI has been framed as an engine of economic prosperity and progress, with little recognition of the human suffering and ecological destruction it perpetuates. Yet, technology does not exist in a vacuum; rather, it reflects the political, economic, and ethical choices of those who develop and control it. AI does not have to function as a tool of digital colonialism—but unless its structural inequalities are addressed, that is exactly what it will remain.
AI’s future should not be built on the backs of exploited workers, poisoned environments, and deepened global inequality. Instead, it must be designed as a truly just and sustainable technology, where its benefits are equitably shared, its costs are fairly distributed, and its governance prioritizes human dignity and planetary survival. This is not a technological challenge—it is a moral and political one. Dismantling AI colonialism requires a fundamental rethinking of who AI serves, who profits from it, and who pays the price for its expansion. It is time for governments, institutions, and civil society to demand accountability—to reject an extractive AI economy and build one that serves humanity, not just the elite few. A future where AI is truly ethical, sustainable, and just is possible – only if we demand it.
References
Cho, R. (2023). AI’s Growing Carbon Footprint. Columbia Climate School. Retrieved from
European Commission. (2024). The AI Act: The First Ever Legal Framework on AI. Retrieved from
Fairwork Project. (2023). Fairwork Cloudwork Ratings 2023: Work in the Planetary Labour Market. Retrieved from
Hao, K. (2019). Training a Single AI Model Can Emit as Much Carbon as Five Cars in Their Lifetimes. MIT Technology Review. Retrieved from
Haskins, C. (2024). The Low-Paid Humans Behind AI’s Smarts Ask Biden to Free Them from ‘Modern Day Slavery’. Wired. Retrieved from
International Labour Organization (ILO). (2023). Generative AI and Jobs: A Global Analysis of Potential Effects on Job Quantity and Quality. Retrieved from
Jones, E., & Easterday, B. (2022). Artificial Intelligence’s Environmental Costs and Promise. Council on Foreign Relations. Retrieved from
Mbembe, A. (2019). Necropolitics. Duke University Press.
Muldoon, J., Cant, C., Graham, M., & Ustek Spilda, F. (2023). The Poverty of Ethical AI: Impact Sourcing and AI Supply Chains. AI & Society. Retrieved from
Oxford Insights. (2023). Government AI Readiness Index 2023. Retrieved from
OpenAI. (2019). Energy and Policy Considerations for Deep Learning in AI. Retrieved from
Pogrebna, G. (2024). AI Underpinned by Developing World Tech Worker ‘Slavery’. Asia Times. Retrieved from
Regilme, S.S.F. (2024). Artificial Intelligence Colonialism: Environmental Damage, Labor Exploitation, and Human Rights Crises in the Global South. SAIS Review of International Affairs 44(2), 75-92. https://dx.doi.org/10.1353/sais.2024.a950958.
Rowe, N. (2023). Underage Workers Are Training AI. Wired. Retrieved from
The Guardian. (2024). Ex-Facebook Content Moderator in Kenya Sues Meta Over Poor Working Conditions. Africa News. Retrieved from
United Nations. (2024). General Assembly Adopts Landmark Resolution on Steering Artificial Intelligence Towards Global Good. Retrieved from
Zuboff, S. (2019). The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. PublicAffairs.
Further Reading on E-International Relations
