Decolonizing Artificial Intelligence
- NOTOC**
Decolonizing Artificial Intelligence
Decolonizing artificial intelligence is an effort to identify and transform the colonial power relations embedded in the development, ownership, governance, and use of artificial-intelligence systems. It examines who controls AI infrastructure, whose knowledge is represented in training data, whose labor makes automated systems possible, and which communities receive the benefits or bear the harms.
The movement goes beyond correcting technical bias. It questions the concentration of data, computing power, capital, research institutions, and decision-making authority in a small number of corporations and wealthy countries. It also challenges the assumption that Western scientific, economic, and ethical frameworks should function as universal standards for all societies.
Decolonial approaches call for AI systems grounded in local histories, languages, institutions, cultures, and knowledge traditions. They emphasize community authority, meaningful participation, informed consent, technological sovereignty, equitable benefit sharing, environmental responsibility, and the right of communities to refuse technologies that threaten their autonomy or cultural survival.
Colonial Power in Artificial Intelligence
Artificial intelligence is frequently presented as a new and neutral technological revolution. Decolonial scholarship argues that contemporary AI is instead connected to older structures of empire, racial hierarchy, resource extraction, and economic dependency.
AI systems depend on vast quantities of data, minerals, energy, water, human labor, and computing infrastructure. These resources are often extracted from communities that have little influence over how the resulting technologies are designed or used. Data and profits commonly move toward corporations and institutions in the Global North, while environmental damage, precarious labor, surveillance, and political dependency remain concentrated elsewhere.
This arrangement has been described through concepts such as digital colonialism, data colonialism, algorithmic colonialism, cognitive imperialism, and technocolonialism. Although these terms emphasize different aspects of the problem, they share a concern with the unequal control of technological systems.
Digital colonialism describes the domination created when foreign technology companies control essential platforms, data centers, cloud services, communication systems, and artificial-intelligence tools. Data colonialism focuses on the transformation of human behavior, culture, language, and social relationships into resources that can be collected and monetized. Algorithmic colonialism examines the imposition of automated systems developed in one social setting upon communities with different histories, institutions, and values.
Decolonial AI therefore asks not only whether an algorithm is accurate, but also who defined its purpose, who chose its categories, who owns the infrastructure, who can challenge its decisions, and who benefits from its operation.
Epistemic Justice and Indigenous Knowledge
A central concern of decolonial AI is epistemic justice: fairness in determining whose knowledge is recognized, preserved, and treated as authoritative.
Most dominant AI systems are built from datasets and classifications shaped by Western academic, commercial, and governmental institutions. These systems may treat Western languages, scientific traditions, social categories, and moral assumptions as universal while representing other knowledge systems as incomplete, informal, or unsuitable for computation.
Indigenous and decolonial scholars challenge the idea that intelligence exists primarily within individuals or machines. Many Indigenous knowledge systems understand intelligence as relational and distributed among people, communities, ancestors, lands, animals, plants, waterways, and spiritual worlds. Knowledge is inseparable from responsibility, place, kinship, and reciprocal obligation.
This creates a fundamental tension with AI development based on unrestricted data extraction. Indigenous knowledge cannot always be separated from the relationships, ceremonies, territories, and responsibilities through which it is maintained. Converting it into machine-readable data can remove it from its cultural context and expose it to appropriation or commercial exploitation.
Decolonizing AI requires more than adding Indigenous information to existing models. It means recognizing Indigenous communities as governing authorities with the power to determine whether their knowledge should be collected, how it may be used, where it is stored, who may access it, and how benefits should be distributed.
Indigenous Data Sovereignty
Indigenous data sovereignty is the right of Indigenous peoples to govern data concerning their communities, territories, languages, cultures, histories, and natural resources.
Conventional data-governance systems often focus on individual privacy and assume that information can be owned or transferred by individuals. Indigenous data sovereignty recognizes that some information belongs collectively to a people and may carry obligations extending across generations.
Artificial intelligence creates new risks for Indigenous data. Language recordings, traditional stories, artwork, ecological knowledge, genetic information, maps, photographs, and ceremonial materials may be scraped from the internet or incorporated into commercial systems without meaningful consent. Generative AI can then imitate Indigenous art, reproduce sacred knowledge, or produce culturally inaccurate material while returning no authority or economic benefit to the communities involved.
Community-centered governance models emphasize free, prior, and informed consent, collective benefit, authority to control, responsibility, ethics, and enforceable benefit-sharing agreements. They also recognize refusal as a legitimate form of technological self-determination. Communities should not be required to digitize their cultures or contribute their data merely because researchers or companies consider the information valuable.
Indigenous-led projects demonstrate that AI can support language revitalization, environmental stewardship, education, and cultural preservation when communities control the data, infrastructure, objectives, and economic value of the technology.
Africa, the Global Majority, and Technological Sovereignty
Africa has become a major focus of debates over AI colonialism. The continent supplies data, labor, minerals, markets, and locations for digital infrastructure while remaining highly dependent on foreign cloud providers, technology companies, research institutions, and funding systems.
Many AI systems deployed in African countries are designed elsewhere and trained on data that poorly represents local languages, agricultural conditions, health systems, legal institutions, and cultural practices. Imported technologies may therefore misidentify crops, misunderstand speech, reinforce administrative discrimination, or produce recommendations that are inappropriate for local environments.
Technological sovereignty refers to the capacity of societies to make independent decisions about digital infrastructure, data, research, standards, and public policy. It does not necessarily require complete technological isolation. Instead, it seeks relationships in which governments and communities can negotiate on fair terms and retain meaningful control over essential systems.
African approaches to AI governance increasingly emphasize regional infrastructure, public-interest computing resources, locally controlled datasets, African languages, Indigenous knowledge, open research networks, and participation by affected communities. Ubuntu and other African relational philosophies have also been proposed as alternatives to individualistic models of AI ethics.
Ubuntu emphasizes that human identity and dignity emerge through relationships with others. Applied to AI, it directs attention toward collective well-being, social responsibility, mutual dependence, and the effects of technology on the whole community.
However, open-source technology alone does not guarantee sovereignty. Countries can remain dependent on foreign computing infrastructure, technical standards, financing, foundation models, and specialized expertise even when software code is publicly available. Genuine sovereignty therefore requires long-term investment in education, institutions, energy systems, research capacity, public infrastructure, and democratic governance.
Global AI Governance and Participation
International AI governance is largely shaped by powerful states, technology corporations, research universities, and international organizations based in wealthy countries. Communities in Africa, Latin America, Asia, the Caribbean, the Pacific, and Indigenous territories are frequently invited to comment on policies after the central priorities have already been established.
Decolonizing global AI governance requires more than symbolic representation. Participation must include authority over agendas, institutional rules, funding, standards, enforcement, and the distribution of technological benefits.
Broad labels such as the Global South can be useful for describing shared experiences of colonialism and economic inequality. However, they can also hide important differences between countries, regions, Indigenous nations, languages, political systems, and social groups. Decolonial governance must therefore respond to specific histories and institutions rather than treating most of the world as a single policy category.
Countries outside the dominant technology powers are not merely passive recipients of AI. They can influence international standards, trade rules, data regulation, procurement, certification, human-rights law, environmental policy, and access to computing infrastructure. Regional cooperation can strengthen their bargaining power and reduce dependency on individual corporations or foreign governments.
Language Justice and Decolonizing NLP
Language is one of the clearest examples of inequality within artificial intelligence. Large language models perform best in languages with extensive digital text, commercial investment, standardized writing systems, and strong representation in academic datasets.
Languages described as low-resource are not naturally lacking in value or complexity. Their digital scarcity often results from colonial education systems, historical suppression, limited publishing infrastructure, economic inequality, and decades of technological underinvestment.
AI systems trained primarily on dominant languages can misunderstand local expressions, erase dialect differences, produce culturally inappropriate content, and perform poorly in safety-critical applications. Automated moderation systems may fail to recognize hate speech, political repression, humor, or community terminology in languages such as Swahili, Tamil, Quechua, and regional forms of Arabic.
Generative systems can also narrow linguistic diversity by encouraging users to adopt standardized styles associated with dominant languages. This may accelerate the marginalization of minority languages and reduce the visibility of local forms of expression.
Decolonizing natural-language processing requires sustained collaboration with speaker communities. Communities should help establish research goals, control language data, define evaluation standards, and determine which materials should remain private or restricted. Successful language technologies should strengthen living communities rather than merely preserve linguistic data for outside researchers.
Community-controlled language models, dictionaries, speech-recognition systems, keyboards, translation tools, and educational resources can support revitalization when they are built according to local priorities. The success of these projects should be measured by community benefit and language vitality rather than only by technical performance.
Labor and the Hidden Human Infrastructure of AI
Artificial intelligence is often described as automated, but it depends on extensive human labor. Workers collect, classify, translate, review, moderate, and label the data used to train and evaluate AI systems.
Much of this work is outsourced to countries where wages are lower and labor protections are weaker. Kenya and other African countries have become important centers for data annotation and content moderation. Workers may be required to review violent, abusive, or sexually explicit material while receiving low pay, temporary contracts, limited psychological support, and little information about the companies benefiting from their work.
Complex subcontracting arrangements can conceal the relationship between workers and major technology companies. This makes it difficult to identify responsibility for working conditions, wages, trauma, discrimination, or wrongful dismissal.
Decolonial labor analysis compares these arrangements to older colonial systems in which wealth was accumulated through the extraction of undervalued labor from politically and economically marginalized populations.
A just AI economy would require transparent supply chains, fair wages, stable contracts, collective bargaining rights, mental-health protections, opportunities for advancement, and public accountability. Workers should also receive recognition for their intellectual contributions rather than being treated as interchangeable sources of manual input.
Environmental Extraction and Data Centers
AI has substantial material and environmental requirements. Training and operating large models consumes electricity and water, while manufacturing computing equipment depends on minerals, industrial facilities, transportation networks, and global supply chains.
Data centers are increasingly constructed in regions where governments hope to attract investment and establish technological sovereignty. However, these facilities can compete with nearby communities for water, land, and electricity. The environmental costs may fall most heavily on Indigenous peoples, low-income neighborhoods, and communities already affected by pollution or resource scarcity.
Minerals required for batteries, semiconductors, and electronic equipment are frequently extracted from territories shaped by colonial land policies. Lithium mining in South America, mineral extraction in Africa, and energy-intensive infrastructure projects can threaten water systems, cultural sites, livelihoods, and Indigenous sovereignty.
Decolonial environmental analysis rejects the idea that AI exists in an immaterial digital cloud. Every automated system depends on physical landscapes, energy systems, workers, and ecological relationships.
Technological sovereignty must therefore include environmental justice. Building locally controlled data centers does not constitute liberation when the infrastructure reproduces pollution, displacement, water depletion, or unequal access to electricity. Communities must have authority to evaluate environmental consequences and determine whether projects provide genuine public benefits.
Education, Healthcare, and Public Institutions
AI systems are increasingly used in education, healthcare, social services, employment, policing, and public administration. These applications can reproduce colonial assumptions when they are imported without attention to local histories and institutions.
In education, generative AI may present dominant cultural interpretations as neutral knowledge. Students can receive inaccurate or stereotypical information about their own histories, languages, and communities. Schools may also become dependent on foreign platforms that collect student data and shape teaching practices.
Decolonial education emphasizes pedagogical sovereignty: the ability of educators and communities to determine what knowledge is taught, how learning is evaluated, and which technologies serve local educational goals.
Healthcare AI presents similar concerns. Models developed from populations in wealthy countries may perform poorly when applied to patients with different genetic backgrounds, disease patterns, infrastructure, languages, or cultural understandings of health. Imported ethical frameworks may emphasize individual consent while neglecting family relationships, collective welfare, historical medical exploitation, and community authority.
Decolonizing public-sector AI requires local evidence, participatory design, transparent procurement, independent evaluation, accessible appeal procedures, and the ability to reject systems that create unacceptable risks.
Culture, Copyright, and Creative Justice
Generative AI systems are trained on enormous collections of art, literature, music, photographs, and cultural expressions. These materials may include Indigenous designs, traditional stories, sacred symbols, and works created by communities whose intellectual-property rights are poorly protected by conventional copyright law.
AI can imitate recognizable Indigenous styles without understanding their meaning or respecting restrictions on who may use them. Synthetic cultural products may compete with authentic community-created works, confuse consumers, and weaken the economic position of Indigenous artists.
Postcolonial writers and artists also face pressure from automated systems trained on simplified ideas of cultural authenticity. Algorithms may reward familiar stereotypes while marginalizing work that does not match commercial expectations about a region or identity.
Creative data justice seeks fair systems of ownership, consent, attribution, compensation, and participation. It also recognizes that some cultural knowledge should not be collected, reproduced, or commercialized at all.
Decolonial design treats culture as a living relationship rather than a source of content. Ethical collaboration requires long-term trust, respect for community protocols, shared authority, and recognition that cultural custodians may impose limits extending beyond conventional copyright.
From Extraction to Co-Creation
Decolonial AI proposes a transition from extraction to co-creation. Extractive development begins with the objectives of companies, governments, or researchers and then seeks data, labor, and community acceptance. Co-creation begins with the priorities of the people who will live with the technology.
Meaningful co-creation requires participation throughout the AI lifecycle, including the identification of problems, collection of data, selection of models, definition of success, deployment, monitoring, and decisions about whether a system should continue operating.
Community consultation must involve real decision-making authority. Participation is not meaningful when communities can provide feedback but cannot alter a project, control their data, receive benefits, or withdraw consent.
Decolonial impact assessments can help identify colonial power relations before systems are deployed. Such assessments examine historical context, ownership, infrastructure dependency, knowledge representation, environmental effects, labor conditions, cultural risks, and the distribution of benefits.
Trustworthy AI standards should include decolonial requirements alongside privacy, safety, transparency, and technical reliability. Certification and auditing systems should evaluate whether communities have genuine authority and whether projects reduce rather than reproduce structural inequality.
Principles for Decolonial Artificial Intelligence
Decolonial AI does not offer a single universal model. Its central purpose is to make universal claims accountable to the communities and histories they affect.
Important principles include community self-determination, Indigenous data sovereignty, epistemic justice, linguistic diversity, relational ethics, environmental responsibility, labor rights, local ownership, democratic participation, and equitable benefit sharing.
Decolonial approaches also recognize the right to refusal. Not every social problem requires an AI system, and not every form of knowledge should be converted into data. Communities must retain the power to establish limits, preserve secrecy, reject harmful applications, and pursue non-technological solutions.
AI development should strengthen local institutions rather than create permanent dependency. This requires investment in public education, research networks, computing infrastructure, language resources, environmental protections, labor standards, and accountable governance.
Conclusion
Decolonizing artificial intelligence means confronting the historical and continuing inequalities that shape digital technology. AI is not created only through code. It emerges from social institutions, political choices, labor systems, cultural classifications, natural resources, and unequal global relationships.
Technical efforts to reduce bias are important, but they are insufficient when ownership, infrastructure, and decision-making remain concentrated. A system can become more statistically accurate while continuing to extract data, displace knowledge, exploit workers, consume community resources, or reinforce foreign dependency.
Decolonial AI shifts attention from inclusion within existing systems toward transformation of the systems themselves. It asks whether communities control their knowledge, whether workers share in the value they create, whether technologies respect ecological limits, and whether societies can determine their own technological futures.
The goal is not simply to create more diverse artificial intelligence. It is to build relationships of knowledge and technology based on sovereignty, reciprocity, responsibility, justice, and the recognition that many forms of intelligence already exist.
- TOC**
Foundations and Decolonial Theory
Artificial Intelligence and Epistemic Justice: A Decolonial Turn Through Indigenous Knowledge Systems
| M. Sethy | AI & Society | 2026
Develops a methodology for integrating Indigenous knowledge systems into AI development to challenge epistemic exclusion and cognitive domination.
From AI Colonialism to Co-Creation: Bridging the Global AI Divide
| Ruhi Khan | LSE Media Blog | 2025-07-14
Calls for locally driven co-creation, participatory governance, and fairer distribution of AI’s economic and social benefits.
Decoloniality Impact Assessment for AI
| David Eke and coauthors | AI & Society | 2025
Proposes an assessment method for identifying colonial power relations and decolonial opportunities across the AI lifecycle.
Cognitive Imperialism in Artificial Intelligence
| Y. Ofosu-Asare and coauthors | AI & Society | 2025
Examines how dominant AI systems privilege Western ways of knowing and presents Indigenous knowledge systems as a counterweight to cognitive imperialism.
Abundant Intelligences: Placing AI Within Indigenous Knowledge Frameworks
| Jason Edward Lewis and coauthors | AI & Society | 2025
Rejects the assumption that intelligence is scarce or uniquely computational and situates AI among plural human and more-than-human intelligences.
Artificial Intelligence Colonialism
| Salvador Santino F. Regilme Jr. | Leiden University Scholarly Publications | 2025
Describes AI as a transnational colonial system built through labor exploitation, environmental extraction, and wealth concentration in the Global North.
Decoloniality as an Essential Trustworthy AI Requirement
| Kutoma Wakunuma | Springer | 2025
Argues that trustworthy AI standards should include decolonial benchmarks, community consultation, certification, and mechanisms for ongoing public feedback.
What Does Decolonising AI Really Mean?
| Ameera Kawash | Untold Magazine | 2024-08-14
Discusses care, labor, scalability, cultural extraction, and the difficulty of building decolonial practices into commercial machine learning.
Decolonial AI as Disenclosure
| W. J. T. Mollema | arXiv | 2024
Uses Achille Mbembe’s concept of disenclosure to propose political, ecological, and epistemic approaches for dismantling AI colonialism.
Artificial Intelligence in the Colonial Matrix of Power
| James Muldoon and Boxi A. Wu | Philosophy & Technology | 2023
Connects AI’s global supply chain to historical colonial structures, showing how data, labor, resources, and decision-making power remain unevenly distributed.
Decolonial AI Alignment: Openness, Viśeṣa-Dharma, and Including Excluded Knowledges
| Kush R. Varshney | arXiv | 2023
Argues that AI alignment can reproduce moral absolutism and proposes openness to models, society, and excluded knowledge traditions.
Relational Autonomy as a Means to Counter AI Harms
| Sabelo Mhlambi | Topoi | 2023
Uses relational African philosophy to challenge individualistic AI ethics and strengthen collective responsibility, dignity, and social accountability.
Risk and the Future of AI: Algorithmic Bias, Data Colonialism and Marginalization
| Arora and coauthors | International Journal of Law and Psychiatry | 2023
Reviews how algorithmic bias and data colonialism intensify power imbalances between the Global North and Global South.
Dependency, Data and Decolonisation: A Framework for Decolonial Thinking in Collaborative AI Research
| Dennis Reddyhoff | arXiv | 2022
Examines extractive academic partnerships and proposes decolonial practices for data empowerment and international AI research collaboration.
Can Artificial Intelligence Be Decolonized?
| Rachel Adams | Interdisciplinary Science Reviews | 2021
Explores whether AI can be separated from colonial logics of race, knowledge, development, and technological authority.
Tech Colonialism Today
| Paola Ricaurte | Data & Society | 2020-02-25
Explains how digital infrastructures reproduce colonial power while highlighting community data practices that resist extractive technological systems.
Artificial Intelligence and Indigenous Perspectives
| Jason Edward Lewis and coauthors | ACM FAccT | 2020-02-07
Argues for a relational shift in AI that respects Indigenous concepts of kinship, responsibility, agency, and collective control.
Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence
| Shakir Mohamed, Marie-Therese Png, and William Isaac | Philosophy & Technology | 2020
Introduces decolonial theory as a practical framework for examining power, value systems, extractivism, and unequal social relations throughout AI research and deployment.
Algorithmic Colonization of Africa
| Abeba Birhane | SCRIPTed | 2020
Compares corporate AI expansion in Africa with colonial domination and warns against imported systems that control data, institutions, and public life.
Data Colonialism: Decolonial Gestures of Storytelling
| Tiara Roxanne | Data & Society | 2020
Connects data extraction with settler colonialism and presents storytelling as a method for restoring Indigenous memory, identity, and agency.
Fay Gale Memorial Lecture: Decolonising Artificial Intelligence?
| Genevieve Bell | Australian National University | 2018
Uses decolonization to disrupt the conventional history of AI and recover alternative technological genealogies, cultures, and epistemologies.
Governance, Sovereignty, and Global Power
Advancing Regional Digital Sovereignty: Policy Responses to AI-Driven Data Colonialism in Africa
| On Policy Africa | On Policy | 2026-06-16
Outlines regional policies for retaining African data value, building shared infrastructure, and reducing dependence on foreign AI providers.
Mapping AI Ethics in Africa: Global Principles and African Realities
| CIGI researchers | Centre for International Governance Innovation | 2026-04-07
Finds that African strategies often endorse global ethics principles while giving insufficient attention to sovereignty, infrastructure dependence, participation, and colonial history.
Toward Self-Determined AI Development in Africa
| Data & Society researchers | Data & Society | 2026-02-25
Draws lessons from earlier development interventions to support African ownership, accountability, and self-determination in AI policy and infrastructure.
The Global Majority in International AI Governance
| Chinasa T. Okolo and Mubarak Raji | arXiv | 2026-01-23
Examines the global AI divide and proposes resource redistribution, institutional reform, and meaningful Global Majority participation in international governance.
Decolonizing the Governance of Artificial Intelligence in Africa
| Jason O. Effoduh | Science and Public Policy | 2026
Advances a governance model grounded in epistemic sovereignty, participatory legitimacy, Ubuntu, and plural African knowledge systems.
High-Stakes Decision-Making and the Future of AI Governance
| E. Kavanagh and coauthors | Journal of Decision Systems | 2026
Critiques globally concentrated AI governance and argues that high-stakes systems require broader participation from communities outside dominant technology powers.
Is Africa Being Recolonized in the Era of Technological Advancement?
| D. O. Okocha and coauthors | SAGE Journals | 2026
Uses expert interviews and digital-colonialism theory to assess whether AI is weakening African sovereignty and reinforcing external control.
The Open-Source Paradox: Africa’s Digital Sovereignty
| O. A. Shonubi and coauthors | AI Magazine | 2026
Asks whether open-source AI provides genuine autonomy or creates new dependencies on foreign compute, models, funding, and technical standards.
Moving Beyond the Term “Global South” in AI Ethics and Policy
| Stanford HAI researchers | Stanford Institute for Human-Centered AI | 2025-11-19
Warns that broad geographic labels can hide local differences and urges AI policy grounded in specific histories, institutions, and power relations.
Digital Sovereignty and Data Colonialism: Shaping a Just Digital Order for the Global South
| Marcus Vinícius de Freitas | Policy Center for the New South | 2025-10
Presents digital colonialism as a structural challenge and recommends coordinated sovereignty strategies for data, infrastructure, regulation, and innovation.
Algorithmic Colonialism
| Mohamed Hssaini | Politikon | 2025
Analyzes Kenya’s dependence on foreign AI systems and proposes a Pan-African sovereignty agency, public models, and cultural data libraries.
AI, Global Governance, and Digital Sovereignty
| Swati Srivastava and Justin Bullock | arXiv | 2024-10-23
Explains how AI redistributes power among states, corporations, and international institutions while reshaping competing ideas of sovereignty.
Decolonizing Global AI Governance: Assessment of the State of Decolonization in Sub-Saharan Africa
| Gizachew Ayana and coauthors | Royal Society Open Science | 2024
Evaluates African representation, institutions, strategies, and regulatory capacity to measure progress toward decolonized AI governance.
Artificial Intelligence, Digital Colonialism, and the Implications for Africa’s Future Development
| A. O. Salami | Data & Policy | 2024
Examines data exploitation and external technology control in Africa and argues for African ownership of digital development.
Can Global South Countries Shape Catastrophic-Risk AI Governance?
| Cecil Abungu, Michelle Malonza, and Sumaya Nur Adan | arXiv | 2023-12-07
Challenges assumptions that Global South states are merely bystanders and identifies strategic roles in international AI governance.
Critical Roles of Global South Stakeholders in AI Governance
| Marie-Therese Png | ACM FAccT | 2022
Maps decolonial AI-governance work from the Global South and identifies roles in trade law, standards, certification, and human rights.
Artificial Intelligence in the Global South: Potential and Risks
| P. J. Wall, Deepak Saxena, and Suzana Brown | arXiv | 2021-08-23
Reviews opportunities and harms when AI designed in the Global North is deployed in different political, cultural, and institutional environments.
Parables of AI in/from the Global South
| Ranjit Singh and Rigoberto Lara Guzmán | Data & Society | 2021-07-13
Frames storytelling as a way to theorize uneven everyday experiences with automated systems beyond dominant Western case studies.
Mapping AI in the Global South
| Ranjit Singh | Data & Society | 2021-01-26
Builds a vocabulary for understanding how digital IDs and AI are adopted, contested, and governed across diverse Global South settings.
Africa and Global Majority Perspectives
AI Is Ushering in a New Era of Colonialism
| Scott Rosenberg and Axios staff | Axios | 2026-06-04
Explores how Western datasets, values, and corporate extraction can flatten cultural difference and marginalize oral and Indigenous knowledge.
From Blooms to Bytes: Will Kenya’s Data Center Boom Repeat the Greenhouse Effect?
| Global Voices contributor | Global Voices | 2026-05-08
Compares Kenya’s AI infrastructure rush with earlier extractive development models and questions who bears the environmental and social costs.
The Human Cost of the Data Center Push in Latin America
| Global Voices contributor | Global Voices | 2026-04-29
Investigates how AI data centers place energy, water, and land pressures on communities while governments promote them as development.
The Hidden Cost of AI: Digital Colonialism and the Global South
| WACC Global | WACC Global | 2026-04-29
Connects AI’s invisible data labor and infrastructure to continuing patterns of extraction, profit concentration, and unequal development.
Western AI Models “Fail Spectacularly” in Farms and Forests Abroad
| Rina Chandran | Rest of World | 2026-03-12
Shows how models trained on Western datasets misidentify crops and ecosystems, demonstrating the importance of local data and community expertise.
AI Ethical Challenges: A Perspective of Developers in Postcolonial Countries
| I. Abraham and coauthors | Information Technology & People | 2026-02-11
Examines West African developers’ experiences with data scarcity, imported standards, infrastructure limits, and culturally mismatched ethical frameworks.
Decolonising AI: A Pan-African Collaboration
| Pan-African research team | ResearchGate | 2026-02-06
Presents a multilingual, values-driven approach to AI in higher education based on South-South-North co-creation and Indigenous African voices.
AI Ethics in Postcolonial Contexts: A Critical Synthesis of Power, Dependency, and Local Agency
| I. Abraham | AI & Society | 2026
Identifies epistemic templating, infrastructural lock-in, governance transfer, and labor opacity as mechanisms through which colonial power operates in AI.
Colonialism Didn’t Die, It Logged In
| International Affairs Forum contributor | International Affairs Forum | 2026
Argues that AI updates colonial domination through data harvesting, digital labor, and platform control rather than territorial rule.
Seeking Voices on AI From the Global South
| ECPD contributors | European Center for Peace and Development | 2026
Calls for linguistic inclusion, locally grounded research, and greater Global South authority in defining AI problems and solutions.
The Future of Africa: Toward Technological Sovereignty or New Dependency?
| Valdai Discussion Club contributors | Valdai Discussion Club | 2025-11-14
Reviews African efforts to build research networks, open knowledge, local infrastructure, and AI capabilities outside Big Tech dependence.
The Mongolian Startup Defying Big Tech With Its Own LLM
| Viola Zhou | Rest of World | 2025-08-22
Profiles a locally controlled Mongolian model designed to preserve language, culture, and technological sovereignty despite limited compute.
Google and Perplexity Give Free AI Search to Win India Users
| Rest of World staff | Rest of World | 2025-08-07
Examines how subsidized AI services can create data dependence, linguistic mismatch, and a new form of platform colonialism.
What Will the AI Revolution Mean for the Global South?
| Krystal Maughan | The Guardian | 2025-08-03
Questions whether AI democratization is credible while compute, conferences, research power, and economic benefits remain concentrated in wealthy countries.
African Data Ethics: A Discursive Framework for Black Decolonial Data Science
| Teanna Barrett and coauthors | arXiv | 2025-02-22
Develops seven principles centered on community, self-determination, African institutions, education, and resistance to anti-Black algorithmic colonialism.
Toward a Decolonial Framework for Communicating AI in Africa
| G. A. Ooko | SAGE Journals | 2025
Calls for African governments and communicators to center African languages, knowledge systems, innovation, and policymaking authority.
Ethics of AI in Africa: Interrogating the Role of Ubuntu
| K. Yilma | Ethics and Information Technology | 2025
Assesses whether Ubuntu and African governance initiatives meaningfully challenge epistemic injustice in mainstream AI ethics.
Global South Must Remain Vigilant Against Technological Colonialism
| Opinion contributor | IDCPC | 2024-12-25
Warns that technological dependency can constrain innovation and reproduce unequal economic and intellectual relations.
Decolonizing LLMs: An Ethnographic Framework for AI in African Contexts
| EPIC research team | EPIC | 2024
Uses research from Ethiopia, Ghana, Kenya, Nigeria, and South Africa to examine cultural mismatch, political tensions, and local adaptation of LLMs.
Are Emerging Technologies Colonial?
| University of Cape Town EthicsLab | UCT EthicsLab | 2023-11-29
Summarizes a discussion of AI, data, and digital colonialism and the political importance of naming technological power relations.
Indigenous Data Sovereignty and Knowledge
In AI Race, Indigenous Values Could Guide Environmental Ethics
| Mongabay staff | Mongabay | 2026-07-07
Examines how Navajo and Māori concepts can reshape AI’s treatment of land, ecological responsibility, and more-than-human relations.
War, Climate Change and AI: What’s at Stake at the UN Indigenous Forum
| Associated Press | AP News | 2026-04-20
Reports Indigenous concerns about AI exploitation, digital extractivism, data sovereignty, land, health, and cultural survival.
A Framework for Kara-Kichwa Data Sovereignty in Latin America and the Caribbean
| WariNkwi K. Flores and coauthors | arXiv | 2026-01-10
Presents an Indigenous legal and relational framework for governing data as ancestral memory rather than a freely extractable commodity.
Toward a Decolonial AI Sovereignty Model for Indigenous Education
| M. F. Mbah and coauthors | Cogent Education | 2026
Proposes a model grounded in Indigenous data sovereignty, relational ethics, refusal, and community authority over generative AI.
Indigenous Ethics and Artificial Intelligence
| M. Maldonado | Discover Artificial Intelligence | 2026
Uses Andean reciprocity to challenge linear Western assumptions about exchange, time, economy, and responsibility in AI ethics.
Data Colonialism and Indigenous Languages in AI
| J. C. Y. Kwok and coauthors | AI & Society | 2026
Warns that Indigenous linguistic data can become raw material for global AI systems without consent, control, or community benefit.
Preventing AI Extractivism: Braiding Indigenous Data Justice With Access and Benefit Sharing
| M. Schulz and coauthors | Tilburg University Research Portal | 2026
Adapts prior-informed-consent and benefit-sharing principles to AI’s extraction of Indigenous linguistic, biometric, ecological, and geospatial data.
Upholding Indigenous Data Sovereignty in the AI Race
| Indigenomics Institute | Indigenomics | 2026
Presents Indigenous data governance as a path from speed and control toward wisdom, relationship, responsibility, and shared benefit.
AI and Indigenous Data Sovereignty: Knowing, Engaging, and Governing
| Special issue editors | Somatechnics | 2025-12-18
Introduces scholarship on Indigenous control over AI-related data, knowledge representation, engagement, and technological decision-making.
AI Reflections: Indigenous Data Sovereignty and Artificial Intelligence
| UBC Indigenous Initiatives | University of British Columbia | 2025-11-19
Explains why Indigenous rights to control data remain central when communities evaluate, build, or reject AI systems.
Indigenous Knowledge Systems and Artificial Intelligence
| Society and AI contributors | Society and AI | 2025-09-14
Proposes co-design, Indigenous data sovereignty, and enforceable benefit-sharing as minimum conditions for ethical AI partnerships.
Calls to Protect Indigenous Intellectual Property From AI Cultural Theft
| ABC News reporters | ABC News Australia | 2025-08-22
Reports warnings that generative AI can scrape, imitate, and commercialize First Nations culture without consent or compensation.
Indigenous Scientists Are Fighting to Protect Their Data—and Their Culture
| Justine Calma | The Verge | 2025-05-12
Profiles Indigenous researchers building governance and storage systems to protect cultural, environmental, and scientific data.
Ancestral Intelligence: How Indigenous Knowledge Informs AI and Data Ethics
| Indigenous Climate Action contributors | Indigenous Climate Action | 2025-04-22
Shows how Indigenous relationality and stewardship challenge extractive AI systems and the structural erasure of Indigenous voices.
Indigenous Data Sovereignty: A Catalyst for Ethical AI in Business
| V. Rana | Business & Society | 2025
Frames Indigenous data sovereignty as a business-ethics response to digital colonialism, exploitation, and misrepresentation.
Indigenous Peoples and Artificial Intelligence: Respect and Consent
| FSC Indigenous Foundation | FSC Indigenous Foundation | 2025
Argues that free, prior, and informed consent must guide the collection and use of Indigenous data and knowledge in AI.
Indigenous Peoples and Artificial Intelligence: A Systematic Review
| M. Perera and coauthors | Big Data & Society | 2025
Reviews research on Indigenous knowledge systems and AI, including risks of appropriation, exclusion, and technological dependency.
Decolonizing Artificial Intelligence: Indigenous Knowledge Systems and Relational Ethics
| S. Khurana | Journal of Critical AI Studies | 2025
Foregrounds land-based intelligence, spirituality, relationality, and responsibility as alternatives to universalist and extractive AI logics.
Peter-Lucas Jones and Māori-Controlled Language AI
Profiles Te Hiku Media’s community-owned Māori speech technology and its insistence that Indigenous peoples retain control over language data and economic value.
AI: A New Revolution or the New Colonizer for Indigenous Peoples?
| Stanford Arcade contributor | Stanford Humanities Center | 2024
Questions whether AI will support Indigenous language and culture or accelerate homogenization, appropriation, and loss of community control.
New Guidelines for Indigenous Data Sovereignty in Artificial Intelligence
| UNESCO | UNESCO | 2023-12-11
Introduces regional guidance for culturally sensitive AI and the data sovereignty of Indigenous peoples in Latin America and the Caribbean.
Artificial Intelligence, Copyright and Indigenous Culture
| Terri Janke and Company | Terri Janke and Company | 2023-11-30
Explains how generative AI can create inauthentic Aboriginal-style art and undermine Indigenous Cultural and Intellectual Property.
Leveraging UNESCO Instruments for Ethical Generative AI Use of Indigenous Data
| UNESCO | UNESCO | 2023-11-08
Applies international ethical instruments to cultural heritage digitization, language revitalization, consent, and Indigenous data protection.
In Consideration of Indigenous Data Sovereignty: Data Mining as a Colonial Practice
| Jennafer Shae Roberts and Laura N. Montoya | arXiv | 2023-09-19
Applies the CARE Principles to AI and data mining and centers Indigenous authority, collective benefit, responsibility, and ethics.
Decolonisation, Global Data Law, and Indigenous Data Sovereignty
| Jennafer Shae Roberts and Laura N. Montoya | arXiv | 2022-07-28
Examines legal and economic incentives for protecting Indigenous cultures, ecosystems, data access, privacy, and rights in the Global South.
Language Justice and Decolonizing NLP
AI-Driven Media and the Reclamation of African Linguistic Heritage
| K. Aiseng and coauthors | SAGE Journals | 2026
Studies how algorithmic visibility affects Setswana, Tshivenda, and Xitsonga and how AI media might support linguistic reclamation.
Large Language Models Are Biased—Local Initiatives Are Fighting for Change
| Laura Vargas-Parada | Nature | 2025-11-27
Profiles local efforts to build models that better serve non-English speakers and culturally diverse communities.
AI Diffusion in Low-Resource Language Countries
| Amit Misra and coauthors | arXiv | 2025-11-04
Links weak language support to lower AI adoption and demonstrates that linguistic accessibility is an independent dimension of the digital divide.
Generative AI and Low-Resource Languages: The Quest for Inclusion
| CTG contributors | Hypotheses | 2025-10-23
Reviews unequal model performance and safety in low-resource languages and the need for linguistic justice in generative AI.
Invisible Languages of the LLM Universe
| Research team | arXiv | 2025-10-13
Argues that the label low-resource language hides historical underinvestment and colonial power relations that produced digital scarcity.
Mind the Language Gap: LLM Development in Low-Resource Contexts
| Stanford HAI researchers | Stanford HAI | 2025-04-22
Maps barriers in data, compute, ownership, evaluation, and funding and recommends community-driven language-model development.
Redefining Technology for Indigenous Languages
| Silvia Fernandez-Sabido and Laura Peniche-Sabido | arXiv | 2025-04-02
Finds that community-developed technologies can strengthen language vitality while externally imposed tools may deepen marginalization.
The Shrinking Landscape of Linguistic Diversity in the Age of LLMs
| Zhivar Sourati and coauthors | arXiv | 2025-02-16
Finds that LLM-assisted writing can homogenize style, suppress difference, and amplify dominant linguistic norms.
Think Outside the Data: Colonial Biases in Moderation for Low-Resource Languages
| Farhana Shahid, Mona Elswah, and Aditya Vashistha | arXiv | 2025-01-23
Shows how English-centered moderation pipelines fail Tamil, Swahili, Maghrebi Arabic, and Quechua and reproduce historical marginalization.
Continuations From European Colonialism to AI: How Languages Are Materialized
| Bettina Migge and coauthors | AI & Society | 2025
Traces continuities between colonial language classification and contemporary computational treatment of linguistic diversity.
Benchmarking Linguistic Diversity of Large Language Models
| Y. Guo and coauthors | Transactions of the ACL | 2025
Evaluates lexical, syntactic, and semantic diversity to measure whether LLMs preserve or narrow human linguistic richness.
The Role of AI in Supporting Indigenous Languages
| Research team | AI and Tech in Behavioral and Social Sciences | 2024-10-01
Synthesizes interviews with speakers, linguists, and technologists about AI’s opportunities and risks for language revitalization.
Harnessing AI to Vitalize Indigenous Languages
| Research team | arXiv | 2024
Examines language revitalization opportunities while emphasizing community ownership, consent, and sovereignty over speech and text data.
Tackling Language Modelling Bias in Support of Linguistic Diversity
| G. Bella and coauthors | HAL / ACM FAccT | 2024
Connects model bias with the loss of linguistic diversity and proposes methods for more plural language technologies.
Layers of Technology in Pluriversal Design
| G. Koch and coauthors | CoDesign | 2024
Uses minority-language technology to show that decolonization requires institutional and political change beyond value-sensitive design.
Using AI to ‘Decolonise’ Language
| Global Voices contributor | Global Voices | 2023-03-07
Surveys civil-society projects using AI to strengthen marginalized languages while questioning control, access, and technological dependence.
Decolonizing NLP for “Low-Resource Languages”
Applies African decolonial ethics to NLP and challenges how researchers define resources, data, and development.
Decolonizing Language Resources in the Human-Machine Era
| Research team | ResearchGate | 2023
Critiques universalist assumptions in language-resource development and asks who benefits from large-scale NLP infrastructure.
Decolonising Speech and Language Technology
| Steven Bird | COLING / ACL Anthology | 2020
Reviews colonizing practices in language technology and proposes community-based methods that support Indigenous language vitality.
Decolonizing Minority Language Technology
| Digital Language Diversity Project | Internet Languages | 2020
Argues that language technologies should be created through close collaboration with speaker communities rather than imposed by outside institutions.
Labor, Extraction, and Environmental Justice
The Environmental Cost of Digital Sovereignty
| Muntaser Syed and coauthors | arXiv | 2026-07-15
Models water, energy, and emissions pressures from sovereign AI infrastructure in the UAE, Bangladesh, India, and Africa.
Pollution From xAI Power Project Hits Black Communities Hardest
| Reuters | Reuters | 2026-07-14
Investigates unpermitted gas turbines powering an AI data center and their disproportionate health impacts on historically Black communities.
Trinidad and Tobago Data Center Deals Raise Resource Concerns
| Associated Press | AP News | 2026-07-12
Reports on planned AI data centers in a country facing water shortages, energy pressures, and questions about development benefits.
Indigenous Leaders Warn AI Boom Repeats Patterns of Extraction
| Grist and Indigenous News Alliance | Truthout | 2026-04-26
Reports that AI can aid land stewardship while its data centers and mineral demand reproduce extractive harms on Indigenous territories.
How AI Hype Masks the Exploitation of African Workers
| Marché Arends and Kathryn Cleary | Tech Policy Press | 2026-03-25
Argues that innovation narratives conceal the appropriation of African labor, skills, and knowledge within global AI supply chains.
Artificial Intelligence and the New Colonialism of Climate Data in the Global South
| PARI researchers | Public Affairs Research Institute | 2025-12-05
Argues that standardized AI climate models can displace local knowledge and transfer authority over vulnerability and adaptation.
The Hidden Kenyan Workers Training China’s AI Models
| Rest of World reporters | Rest of World | 2025-12-04
Investigates informal labor networks, weak protections, and digital-colonial concerns surrounding Kenyan workers training Chinese AI systems.
“Magic” AI Is Exploiting Data Labour in the Global South
| RESET contributors | RESET | 2025-11-19
Connects precarious data work with data-center energy and water consumption while documenting worker resistance.
The Perilous Future of AI Work in the Global South
| LSE Media contributors | LSE Media Blog | 2025-11-14
Examines how workers remain confined to invisible, low-paid tasks while AI profits and ownership accumulate elsewhere.
Reimagining the Future of Data and AI Labor in the Global South
| Michelle Du and Chinasa T. Okolo | Brookings Institution | 2025-10-07
Recommends labor protections, transparency, mental-health support, collective bargaining, and culturally competent moderation tools for data workers.
Kenya’s Youth Face Exploitation in ‘AI Sweatshops’
| Anadolu Agency | Anadolu Agency | 2025-08-22
Reports low pay, short contracts, trauma, and colonial comparisons among Kenyan workers training systems for global technology companies.
The Real Cost of AI Is Being Paid in Deserts Far From Silicon Valley
| Karen Hao | Rest of World | 2025-05-26
Shows how AI-linked mineral and water extraction affects Indigenous communities in Chile’s Atacama region.
Empire of AI Explores Global Costs to Marginalized Workers
| Michelle Kim | Rest of World | 2025-05-19
Discusses how frontier AI companies can pursue innovation without treating workers and communities as disposable inputs.
How Big Tech Hides Its Outsourced African Workforce
| Rest of World investigators | Rest of World | 2025-04-21
Maps the opaque subcontracting networks that separate African data workers from the global companies benefiting from their labor.
African Digital Colonialism Is the New Face of Worker Exploitation
| ICTworks contributor | ICTworks | 2025-04-17
Describes how African labor, data, and resources are extracted without equitable returns or durable local capacity.
In Kenya’s Slums, They’re Doing Our Digital Dirty Work
| Isobel Cockerell | Coda Story | 2025-03-31
Profiles underpaid content moderators and annotators whose hidden labor supports Silicon Valley AI products.
Data Labelling as Digital Taylorism
| Researcher | Toplum ve Kültür Araştırmaları Dergisi | 2025-02-25
Analyzes repetitive data-labeling work as a tightly controlled labor process concentrated in precarious Global South employment.
AI Accelerates Ecological Disaster and Reinforces Injustice
| Collective of researchers and activists | Le Monde | 2025-02-06
Links AI expansion to energy use, labor exploitation, neocolonial dependency, austerity, and concentrated corporate power.
Data Workers in AI: A New Frontier of Labour Exploitation
Describes unfair pay, opaque contracts, weak representation, and limited upward mobility for data workers in Kenya and the Global South.
Moving Toward Truly Responsible AI Development in the Global AI Market
| Brookings researchers | Brookings Institution | 2024-10-24
Reviews low wages, precarious work, psychological harm, and weak regulation affecting data annotators in developing countries.
Global Inequalities in the Production of Artificial Intelligence
| Antonio A. Casilli and coauthors | arXiv | 2024-10-18
Compares data work in Venezuela, Brazil, Madagascar, and France and finds supply chains that reproduce colonial economic dependencies.
Greening Extractivism: Justifying AI Supply Chains in Canada
| Conference researcher | EASST-4S | 2024-07-19
Shows how green AI narratives can rely on colonial views of land and suppress Indigenous understandings of life and ecology.
James Muldoon, Mark Graham and Callum Cant: AI Feeds Off Human Work
| Zoë Corbyn | The Guardian | 2024-07-06
Discusses the extraction machine behind AI and the labor of data workers in Kenya, Uganda, warehouses, and platform supply chains.
An Elemental Ethics for Artificial Intelligence: Water as Resistance
| Sebastian Lehuede | arXiv | 2024-03-11
Builds an AI ethics from Chilean and Indigenous resistance to water-intensive data centers, lithium mining, pollution, and e-waste.
Native Communities Confront Lithium Mining’s Threats to Water and Culture
| Associated Press | AP News | 2024
Documents Indigenous resistance to lithium extraction in South America, an upstream resource issue for digital and AI infrastructure.
Education, Health, Culture, and Design
Towards Participatory AI With ovaHimba and San People
| Research team | ACM | 2026-06-14
Explores co-created AI design with African Indigenous communities rather than imposing externally defined problems and solutions.
Pedagogical Sovereignty and Its Contradictions
| Siham Rebbah | Postdigital Science and Education | 2026-06-06
Critiques decolonial AI in education and examines the tensions between technological sovereignty, pedagogy, infrastructure, and institutional dependence.
The Serpent in the Code: AI and Postcolonial Literary Authenticity
| Global Voices contributor | Global Voices | 2026-06-02
Uses a literary controversy to examine how AI reproduces formulas of authenticity imposed on Caribbean and postcolonial writing.
Entwining Technology With Indigenous Knowledges
| Research team | ACM | 2026-04-13
Examines design practices that treat Indigenous knowledge as living relationships rather than extractable content.
Artificial Intelligence and Traditional Cultural Expressions
| Becky Montesdeoca | ResearchGate | 2026-03-10
Analyzes legal protections for traditional cultural expressions when AI systems copy, transform, and commercialize cultural materials.
Artificial Intelligence, Copyright and Indigenous Culture: The New Frontier
| Terri Janke and Company | Understorey | 2026-01-07
Summarizes ownership, consent, and cultural-protection issues raised by generative AI in Australia.
Algorithmic Dependence and Digital Colonialism in Global South Education
| S. A. Ahmed and coauthors | Frontiers in Education | 2026
Develops a framework covering data, infrastructure, epistemic, and governance colonialism in education systems.
What Does AI Ethics Look Like if We Take Decolonization Seriously?
| Selena Nemorin and coauthors | AI & Society | 2026
Reconsiders AI ethics through colonial history, social institutions, power, and the cultural conditions in which automated systems operate.
Whose Data, Whose Culture? AI and Cultural Knowledge in Education
| Y. Ismaili and coauthors | Globalisation, Societies and Education | 2026
Challenges AI’s authority to produce cultural knowledge and compares how model origins shape educational representations.
Human-Centered AI and African Knowledge Systems in HCI
Integrates African Indigenous knowledge systems into human-centered AI and HCI design.
Relational AI in Education
| Research team | arXiv | 2026
Develops educational AI around reciprocity, participation, consent, and resistance to settler-colonial data extraction.
Decolonial Literary Resistance and Algorithmic Colonialism
| M. Rahmatullah | Journal of Postcolonial Writing | 2026
Reads Indigenous futurism as resistance to algorithmic colonialism and computational control over identity and imagination.
Artificial Intelligence, Technocolonialism and Decolonisation in African Social Work
| African social work scholars | African Social Work | 2025-09-26
Calls for social-work education that critically evaluates AI, centers Indigenous knowledge, and participates in technology design.
Decolonizing AI Ethics in Africa’s Healthcare
| M. K. Grancia and coauthors | Discover Artificial Intelligence | 2025
Critiques imported healthcare-AI ethics and proposes principles rooted in African contexts, values, justice, and inclusion.
Creative Data Justice: A Decolonial and Indigenous Approach
| Payal Arora and coauthors | Information, Communication & Society | 2025
Reimagines creative value, ownership, and participation in the age of AI-enabled cultural production.
Ethical Co-Development of AI Applications With Indigenous Communities
| Claudio S. Pinhanez and coauthors | ACM | 2025
Provides guidance for researchers and practitioners co-developing data-intensive systems with Indigenous communities.
Colonial Structures in AI: A Latin American Decolonial Literature Analysis
| H. Correa Lucero and coauthors | AI & Society | 2025
Synthesizes Latin American scholarship on structural power, data extraction, dependency, and decolonial alternatives in AI.
Towards Decolonising the Ethics of AI in Education
| Selena Nemorin | Globalisation, Societies and Education | 2024
Frames decolonizing AI ethics as resistance to epistemic violence in educational technology.
A Decolonial Approach to AI in Higher Education Teaching and Learning
| Michalinos Zembylas | Learning, Media and Technology | 2023
Proposes educational strategies that resist algorithmic coloniality and question dominant assumptions in AI ethics.
Decolonising AI: A Transfeminist Approach to Data and Social Justice
| Rising Voices and APC | Rising Voices | 2020-01-22
Connects AI, human rights, gender, and social justice through country perspectives from marginalized communities.
