In this interview with Ulises Ali Mejias, Professor at the State University of New York and co-author of The Costs of Connection and Data Grab, we explore data colonialism and its implications in the age of artificial intelligence. Drawing on the historical continuities between colonialism and capitalism, Mejias explains how data has become a new resource for power, epistemic colonialism and social control. He also discusses digital sovereignty and the future of AI, offering a critical perspective on how societies in the Global South and the Muslim world can rethink technology, governance and the digital future.
1. What exactly do you mean by data colonialism? Why do you regard it as an actual form of colonialism rather than as a metaphor?
Prof. Ulises Ali Mejías: The concept of data colonialism offers an analytical framework for understanding what is taking place in data-driven processes more broadly. In other words, the issue is not confined to social media or artificial intelligence; the aim is to make sense of the data relations that emerge across the various domains of digital technology, from mobile devices to cryptocurrencies. This approach requires that we situate these processes not only within the fifty-year history of networked computing systems, but also within the five-hundred-year history of colonialism.
Our central argument is the following: the process we confront today does not rest solely upon the seizure of land, as it did during the periods in which European powers colonised Latin America, Africa, and much of Asia. This time, the matter is less a land grab in the classical sense than the seizure and appropriation of data. Land and labour, of course, retain their importance. Land remains central in the case of data centres, for instance, while labour continues to play a central role when we consider those employed within the gig economy on platforms such as Uber or food-delivery services. In this analysis, however, our principal focus is on how data operates as a new domain of colonisation.
When we speak of data colonialism, we do not mean to imply that we are opposed to data, or that all forms of data are harmful. We define data colonialism as an emerging social order grounded in the continuous appropriation of the data of our lives for the purpose of constructing a new worldview oriented towards two principal ends: the generation of wealth for a small elite, and the construction of new forms of social control. This is precisely what we mean when we say that data colonialism is not a metaphor. It is an emerging social order, and in order to demonstrate this we look to historical continuities.
We do not claim that it appears or operates in exactly the same manner as earlier forms of colonialism. Our argument is rather that, if we trace the continuities that persist throughout colonialism, we can understand why we designate this appropriation of data as a new form of colonialism. The contexts, the methods, the modes of operation, and the ways in which it is experienced by different populations around the world are of course distinct. Yet all forms of colonialism share one fundamental similarity: function.
Colonialism has always operated in the same way: to dispossess, to extract the substance of something, to take what does not belong to you and to seize it unlawfully in order to create wealth for yourself as a coloniser. We therefore attend to the ruptures within colonialism while remaining aware that its legacy is very much still with us. This new form does not replace the older colonialism; it extends and perpetuates that legacy.
To understand all of this, we must also grasp the relationship between colonialism and capitalism. Colonialism and capitalism continue to develop together and to shape one another. To put it simply, without the wealth accumulated through the colonial plantation system, the emergence of the factory and of capitalism might not have been possible. That wealth has been transmitted historically and has continued to shape the world we inhabit today.
2. You argue that data is territory and that our lives are the regions being colonised. How, then, does the rise of artificial intelligence deepen this seizure of territory?
Prof. Ulises Ali Mejías: It is entirely apt to speak of artificial intelligence, because when we began writing about data colonialism, artificial intelligence had not yet attained its present prominence. I believe that artificial intelligence illustrates data colonialism in very clear ways.
In order for artificial intelligence systems to function, extremely large-scale datasets, referred to as “training data”, are required, and a significant proportion of these data is derived from users’ digital activity. In the first phase, artificial intelligence companies gathered a wide range of information and content accessible on the internet and incorporated it into the sphere of private property. Wikipedia entries, social media posts, YouTube content, and the data produced within learning management systems and health platforms may all be considered in this context. This body of knowledge, public in character, was appropriated by corporations and used to train artificial intelligence models.
This situation reveals a significant contradiction with respect to intellectual property and labour relations. Whereas individuals who download content they do not own without authorisation may be accused of piracy, corporations that engage in the same practice on a vastly greater scale frequently encounter no comparable legal or moral sanction. Meta’s training of its Llama model on terabytes of academic articles and books, and Google’s training of its MusicLM system on thousands of hours of musical recordings, may be cited as examples of this process. In most instances, such practices have taken place without licensing procedures, without the permission of content producers, and without any payment.
In the second phase, these models continue to require ever greater quantities of data. Although artificial intelligence systems always carry the potential to err and to hallucinate, their developers maintain that improving these systems depends upon a continuous and steadily increasing supply of data. At this point, the matter is no longer merely the harvesting of Wikipedia entries or online content; it is the real-time capture and processing of our interactions with our computers and telephones, of our driving data, and of our health information.
Unlike the surveillance practices of a decade ago, which were conducted largely for the purposes of targeted advertising, this process is increasingly acquiring a colonial and imperial character. The data collected today can enable corporations to intervene in electoral processes, to erode democratic institutions, to reinforce mechanisms of coercion and control directed at migrants, and to target civilians in wartime. What is at stake here, therefore, is not merely surveillance conducted for commercial ends, but a regime of surveillance that deepens social control and that, in certain cases, can serve the destruction of particular groups of people.
3. From education to migration policy, who pays the heaviest price for data colonialism? In what ways does this system deepen existing inequalities?
Prof. Ulises Ali Mejías: We are witnessing the continuation of legacies and of older injustices that began with colonialism. It is therefore unsurprising that some people should pay a heavier price than others.
What is emerging here is a new kind of “artificial intelligence bureaucracy.” In artificial intelligence, accuracy or precision is not the foremost priority. Even the companies themselves acknowledge that these systems are prone to error and to hallucination. Nevertheless, a new colonial bureaucracy has arisen in which artificial intelligence is permitted to determine who may cross a border, who may receive healthcare or social assistance, and who may gain access to financial services. Whether these decisions are entirely correct is of no real importance, because what is actually intended to be produced is a sense of panic- the feeling that one is being watched and that one may be confronted with the power of the state at any moment.
The deployment of artificial intelligence on the battlefield, in healthcare, and in migration policy delegates the automation of prejudice and inequality to artificial intelligence bureaucrats. People are thereby able to transfer responsibility onto the technological system, “it is not our fault; the artificial intelligence made the decision”, and so to evade accountability.
Education stands out as an area of particular concern in the context of artificial intelligence’s effects. The entry of artificial intelligence into education is producing what is commonly termed “brain rot”: it weakens the cognitive capacities of students and academics alike and erodes the capacity for critical thinking. As the value of the university degree declines, institutions are closing down fields such as the humanities and the arts in order to concentrate on profitable areas.
4. Classical colonialism legitimated itself through the language of modernisation. Today, the great technology corporations arrive bearing promises of digitalisation and development. Consider the enormous investments that the Gulf states are making in artificial intelligence and smart technologies. Where does the fine line lie between progress and the loss of data sovereignty in Muslim-majority countries?
Prof. Ulises Ali Mejías: Colonialism has historically been legitimated through discourses of progress, modernity, and even salvation. Within this discursive framework, the assumption that the coloniser is the subject who best knows what is right, necessary, and beneficial has been continually reproduced. A similar logic can be observed at work in the discourses constructed around artificial intelligence today. Claims of the sort that “we must accept artificial intelligence because it is here to stay and because it will eliminate many fundamental problems, from curing cancer to solving global warming” are contemporary instances of this discourse. Although such promises are often exaggerated and far removed from reality, they are placed in circulation and legitimated on the assumption that the executives of technology companies know best what is good and necessary.
In order to challenge these narratives, a fundamental question must first be posed: to whom does the authority to define progress belong? Does progress mean the accumulation of economic wealth for a narrow elite, or does it mean the opening of everyday life, from the moment we step into a vehicle, for instance, to continuous surveillance and data collection? This question is significant in that it demonstrates that technological development is not merely a technical matter, but also a political, ethical, and social domain of choice.
Unfortunately, the states of countries that identify themselves as Muslim have likewise not performed notably well in this regard. Many of them are adopting a position aligned with the new world order that the West terms “Pax Silica”; they are essentially investing in United States- and Western-centred technology ecosystems and embracing their discourses of progress. Muslim countries ought, however, to focus first upon the following questions: which values are we endorsing when we adopt these narratives? Whose interests does this technological and economic order ultimately serve?
5. In this connection, artificial intelligence systems are trained on the data of particular languages and cultures and reproduce the values of those cultures. Does the absence or distortion of the Islamic intellectual tradition and worldview within this system give rise to a new form of epistemic colonialism? What ought knowledge sovereignty to mean for the Muslim world today?
Prof. Ulises Ali Mejías: All technologies reproduce the epistemic biases of their producers. Today, in Latin America, in Europe, and in the Islamic world, there is a debate about “building models trained on our own local data.” But are such models sufficient to dismantle epistemic colonialism? My answer is that they are not.
If these models are to be used in order to substitute machines for human labour, to put people out of work, to weaken students’ critical thinking, or to construct environmentally polluting mega data centres of the kind we see in the Gulf states, what benefit is there in a “Muslim artificial intelligence model”? For an artificial intelligence model to be national or sovereign merely because it uses local data means nothing at all.
What genuinely binds a technology to Islamic values is not the language in which it is trained, but whether it is so constituted as to seek the greatest social benefit, to uphold egalitarian and democratic models of governance, and to aim at the elimination of suffering and poverty.
Question: You call into question the assumption that being connected is always a positive condition, given that data relations are so deeply woven into the fabric of everyday life that exiting them at the individual level appears all but impossible. In this context, what does the decolonisation of data mean? Should resistance begin primarily at the intellectual and conceptual level, or should it be built through structural and collective interventions?
Prof. Ulises Ali Mejías: Personally, I am not opposed to the condition of being connected; my basic argument is that connectivity must be redefined in accordance with our own terms, our own values, and our own social priorities. Technologies must therefore be built to serve the good of society rather than the interests of corporations.
To decolonise data is to pursue this plan of action at every level. Individual action alone is not sufficient; it is necessary to exert pressure upon governments for legislation and to build collective alliances across communities and across different systems of value. We call this working simultaneously within the system, against the system, and beyond the system. Within the system we must exert pressure through institutional politics; against the system we must engage in activism; and beyond the system we must generate our own solutions.
Our most important weapon here is the imagination. Colonialism colonises the mind as well, and we must therefore also decolonise the mind, transforming ideas and creative works into action.
6. Many countries today have embraced the discourse of data sovereignty. Yet technological infrastructure, cloud services, artificial intelligence models, and digital platforms continue to remain under the control of a handful of global corporations. In your view, what steps should Muslim countries take in order to attain genuine digital independence? Does the solution lie chiefly in producing technology, in legal regulation, or in creating social awareness?
Prof. Ulises Ali Mejías: In a globalised world, no country can produce the whole of its technology on its own, nor can it be fully independent. Sovereignty is a Western construct that promises absolute autonomy. We exist as interdependent collectives. The solution, therefore, is not sovereignty alone but the public governance of technologies. Indeed, public ownership of companies such as Google or Meta would not constitute a solution either, since it would achieve nothing beyond allowing those companies to continue operating as they presently do.
What we require is the establishment of a “Non-Aligned Technology Movement” akin to the Non-Aligned Movement of the Cold War era. The solution is not to replace Western technology with Chinese or Muslim technology. None of us is free until all of us are free. We must build new legal and social frameworks of production that exalt the human capacity for original thought, that strengthen the sense of responsibility and of mutual attachment, and that protect the environment. Islamic values, too, can play a significant role in this movement by forming global alliances that uphold the best of humanity.
