Artificial intelligence has become the newest frontier of global power, extending its old hierarchies into the digital realm. What once relied on territorial conquest now depends on algorithmic reach. Across the Global South, vast amounts of data and digital labor are harvested to train AI systems owned and governed by the Global North. This process, as described by Nick Couldry and Ulises Mejias as data colonialism, turns human experience into raw material for profit and political influence. Yet the issue extends beyond economics. The same systems that extract data also shape thought. Through biased datasets and generative models built on Western epistemologies, AI technologies export cultural norms and rewrite global cognition. As a 2025 report in Modern Diplomacy notes, this “digital dependency” allows a handful of Western corporations to dominate global information systems, mirroring colonial patterns of control. This extraction of data and labor forms the foundation for a subtler kind of empire: digital re-education.

Invisible Labor, Visible Control

Before AI systems can think, they must first learn. And much of this is learned by unchecked harvesting of data and labor from the Global South. This is seen where big tech companies often outsource workers to perform traumatizing tasks to not only benefit from cheap labor, but also avoid jurisdiction on working conditions. In an investigation done by TIME magazine in Kenya, a tech company SAMA made about 19$ million while its workers in Kenya were making $1-$2 per hour as content moderators filtering through violent, and traumatizing videos and photos covering everything from executions to child abuse.

These human injustices mirror a deeper digital dependency. Across Africa, the rise of AI has deepened dependency on foreign technology firms that control data flows and infrastructure. Cambridge Data & Policy (2024) reports that most digital platforms, services, and undersea cables are owned or managed by corporations based in the Global North. This dominance gives them direct power over how information circulates, shaping not only local economies but also cultural and political life. Through “free” or subsidized services like Facebook’s Free Basics and Google Maps, major tech companies harvest user data and behavioral patterns from millions of people, often without meaningful consent. As African Business (2023) and Solon (2017) note, these programs are framed as tools for inclusion but primarily function as systems for large-scale data extraction and profit. The result is a digital ecosystem where users become sources of information rather than participants in its governance.

These same extractive methods also extend into the physical world, where the demand for AI infrastructure is reshaping communities and depleting their resources. The lack of its own users being participants in their own governance can even be seen when examining resource extraction for AI software: their data centers consume vast amounts of water for cooling, further straining supplies in regions already grappling with water scarcity. In Uruguay, mass protests erupted after the country’s worst droughts, linked to Google’s data centers. Meanwhile, indigenous communities in Chile face scarce water and ecological harm from lithium extraction. Yet the most profound impact of this digital extraction is not economic but cognitive, as when we see these systems in effect, the same ones that are harvesting data are also teaching us how to see the world.

From Extraction to Re-Education

Once trained, large AI systems currently not only process data but reproduce the cultural assumptions and hierarchies of the Global North under the guise of technological neutrality.  Every translation, search result, and image suggestion quietly shapes what people see as real, trustworthy, or worth knowing. In Boaventura de Sousa Santos (2024) piece, AI design is rooted in Western epistemologies that define what counts as rational knowledge, while Couldry and Mejias (2019) describe this as data colonialism as a structure that transforms human experience into a resource for global inequality. Generative models rarely include non-Western histories, languages, or worldviews, narrowing what the digital world recognizes as true. The problem isn’t just who builds AI, but what it ends up teaching — spreading Western values and ways of seeing as if they were universal.

This re-education is most visible through language. The arXiv (2024) paper Generative AI and Digital Neocolonialism shows that English-centric systems erase indigenous knowledge, while Mohamed et al. (2020) find that most NLP models ignore non-Western languages altogether. In São Paulo, Google’s Plus Codes privatized mapping data for informal settlements, reducing community control. Similarly, Maori communities have resisted Big Tech attempts to use their language data, safeguarding it to prevent corporations from exploiting te reo Maori and the traditional knowledge embedded within it. By prioritizing certain voices while shutting others out, trains AI not to just reflect bias, but to teach it.

Reclaiming Digital Autonomy

While the reach of AI reflects a widening digital divide, growing movements across the Global South are pushing to reclaim technological sovereignty. Scholars and policymakers are calling for what the LSE Media Blog (2025) terms a shift “from AI colonialism to co-creation,” emphasizing equity, consent, and cultural inclusion in data governance. The African Union’s AI Strategy (2024) and initiatives like UNESCO’s Recommendation on the Ethics of Artificial Intelligence (2023) both outline frameworks for ethical, rights-based development that prioritize local control over data and algorithms. These approaches acknowledge that data sovereignty is inseparable from political sovereignty by investing in indigenous language models, public data infrastructures, and community-owned digital platforms. With this, AI can offer a path more towards autonomy and empowerment rather than a tool of extraction.

Conclusion

As AI systems built on Global South data continue to define global knowledge , the fight for human rights becomes inseparable from the fight for digital sovereignty. Without confronting this new form of re-education, the 21st century risks exchanging colonial empires for algorithmic ones, and the world’s most valuable territory will not be land, but the human mind.

Sofia Nazeer is an undergraduate senior at the School of International Studies at American University.

Works Cited

African Business. (2023). Digital Colonialism and the Data Economy. – https://african.business/2023/06/technology-information/the-risks-of-digital-colonialism-in-africa/

Cantarini, M. (2022). Mapping Inequality: Google Plus Codes in São Paulo. Confins Journal. – https://journals.openedition.org/confins/49040

Cambridge Data & Policy. (2024). Africa’s Dependency on Foreign AI Infrastructure. – https://www.cambridge.org/core/journals/data-and-policy/article/artificial-intelligence-digital-colonialism-and-the-implications-for-africas-future-development/4BD73E9129A9CD9E9301C61CB2401450

Couldry Nick & Ulises Mejias (2019): The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism. Stanford University Press. Link: https://www.sup.org/books/sociology/28816/

Cuthbertson, A. (2023, July 11). Uruguay drought: Water crisis sparks protests over Google data centre. The Guardian. https://www.theguardian.com/world/2023/jul/11/uruguay-drought-water-google-data-center

LSE Media Blog. (2025). From AI Colonialism to Co-Creation. – https://blogs.lse.ac.uk/medialse/2025/04/02/from-ai-colonialism-to-co-creation/

Mohamed, S., Png, M.-T., & Isaac, W. (2020). Decolonial AI: Challenges of Western-Centric NLP. ACM FAccT. – https://dl.acm.org/doi/10.1145/3351095.3372827

Modern Diplomacy. (2025). Digital Dependency and AI Power Structures. – https://moderndiplomacy.eu/2025/03/11/digital-dependency-global-ai-report/

Natural Resources Defense Council. (n.d.). Lithium mining is leaving Chile’s indigenous communities high and dry, literally. NRDC. https://www.nrdc.org/stories/lithium-mining-leaving-chiles-indigenous-communities-high-and-dry-literally

Santos, B. de S. (2024). AI and the Epistemologies of the South. Journal of World-Systems Research, 30(2), 635–645. https://doi.org/10.5195/jwsr.2024.1291

Solon, O. (2017). It's digital colonialism': how Facebook's free internet service has failed its users The Guardian. https://www.theguardian.com/technology/2017/jul/27/facebook-free-basics-developing-markets

TIME Magazine. (2023). Inside the Exploitation of Kenyan Workers Powering ChatGPT. – https://time.com/6247678/openai-chatgpt-kenya-workers/

UNESCO. (2023). Recommendation on the Ethics of Artificial Intelligence. – https://unesdoc.unesco.org/ark:/48223/pf0000381137

WIRED — “Maori are trying to save their language from Big Tech”. WIRED. 2021. Link: https://www.wired.com/story/maori-language-tech/

African Union. (2024). AI Strategy for Africa. – https://au.int/en/ai-strategy

arXiv. (2024). Generative AI and Digital Neocolonialism. – https://arxi