Beyond the digital enclosure: AI, coloniality, and the pursuit of educational sovereignty

This is the author’s accepted manuscript (AAM) of the article published in Learning Futures and Emerging Technologies. The final version of record is available from Emerald Publishing at https://doi.org/10.1108/LFET-01-2026-0004 .

Abstract

Purpose

This conceptual paper aims to examine the rapid adoption of generative AI in public education, arguing that current “readiness” frameworks can mask a transfer of authority from public institutions to private platform firms. It provides a structural framework for understanding digital enclosure, coloniality and educational sovereignty in AI-mediated educational settings.

Design/methodology/approach

The paper synthesizes decolonial theory with work on digital enclosure, platform infrastructure, surveillance capitalism, epistemic justice and educational technology. It introduces the “four lanes of coloniality” as a framework for analyzing institutional dependency under vendor-controlled AI adoption.

Findings

The paper identifies four lanes through which platform power enters education: infrastructure, classification, epistemology and labor. These lanes operate through vendor dependency, automated surveillance, linguistic bias, epistemic gatekeeping and uncompensated repair labor.

Research limitations/implications

This is a conceptual paper focused on closed-weight, cloud-hosted, vendor-controlled AI systems. Empirical studies should test the framework across different institutions, governance models and AI arrangements.

Practical implications

The paper outlines a minimum sovereignty agenda for institutions adopting AI. This includes stronger procurement rules, audit rights, visible system changes, usable data access, appeal processes, recognition of repair labor and meaningful participation by educators and students.

Social implications

Reclaiming educational sovereignty is presented as essential for maintaining education as a public good and preventing public learning from becoming a resource frontier for private capital.

Originality/value

By framing AI adoption as a governance and coloniality problem rather than a mere pedagogical shift, the paper offers a four-lane framework and a practical diagnostic for institutional reflection and action.

Keywords: Artificial intelligence, coloniality, platform governance, digital enclosure, educational sovereignty

Public education in an enclosure

The explosive growth of generative AI in education marks a crisis of governance, well beyond a simple shift in pedagogy. Big Tech has long treated public education as a quarry, open for extraction, arriving with a sense of inevitability while institutional leaders invoke “readiness” and “competitiveness.” As platform firms market modernization, they mask a transfer of authority. The result is a chain of fear: administrators worry about reputational obsolescence, faculty anticipate accusations of misconduct and students brace for a future in which they will not be prepared. Bound by institutional isomorphism, schools implement copy-and-paste policies rather than govern on their own terms (DiMaggio and Powell, 1983).

The enclosure begins with routine institutional gestures that are born from prudence. For example, a committee adds a syllabus statement on disclosure. Or a provost circulates a “responsible use” memo. Or a vendor offers free training and a campus license “for experimentation.” Faculty receive templates, checklists and sample prompts. Administrators can present action without committing substantial resources. These moves shift attention to how the tools are used, conduct, while leaving platform governance untouched. The institution starts to manage risk at the classroom level while it cedes authority at the system level.

These gestures may seem easily justifiable, but they represent what Andrejevic (2007) describes as the digital enclosure: private platforms sequester activities that were previously public or unmediated, and they make surveillance the price of admission. In this arrangement, the institution becomes a site of surveillance capitalism (Zuboff, 2019), where interaction logs become assets for private accumulation. Big Tech retains the source code, the capital and the unilateral right to alter systems that increasingly frame the conditions of learning.

Much has changed over the past decade, yet much of the current discourse treats AI as if it belongs in the same category as calculators or mobile phones. “AI” is a system, not a package of simple products. Calculators were not adopted with cloud contracts, telemetry and content policies that silently govern inquiry. Or with private regimes of ranking and summarization that reshape what is deemed credible knowledge. When schools play along with the analogy, they treat AI as a classroom management problem rather than a political and epistemic shift. That framing makes platform rulemaking appear natural.

These dynamics are visible in ordinary adoption decisions. A university licenses an AI writing assistant through single sign-on, which turns student drafting, revision and inquiry into platform-mediated data. Or an instructor relies on an AI detector after a misconduct concern, which shifts proof burdens onto students whose language does not match dominant academic registers. Or a learning platform adds AI summaries, dashboards or tutoring features through a vendor update, which changes the conditions of teaching before faculty have reviewed the pedagogical assumptions. Each case may appear procedural and innocuous when viewed alone. Taken together, however, they show how authority moves upstream, from educators and students toward platforms that set rules through infrastructure, classification and interface design.

This article provides a critical conceptual perspective and framework-building intervention. It draws from digital enclosure, platformization, surveillance capitalism, coloniality, epistemic justice and educational technology to examine how vendor-controlled AI systems can reorganize educational authority. The aim is not to test an empirical claim, but to name a governance pattern and offer educational sovereignty as a practical diagnostic for institutional decision-making.

Frameworks without agency

AI enters education wrapped in guidance. Frameworks, checklists and training modules promise safety and clarity. Many are well intentioned. Most, however, govern the conduct of teachers and students while leaving platforms beyond accountability. They train compliance at the edge of the system, while the center remains vendor-controlled: updates, telemetry, content rules and ranking behavior. The result is a professional ethic reduced to performance, framed as procedure.

This mismatch produces a predictable colonial pattern: responsibility moves downward through disclosure rules and integrity policies, while authority moves upward through vendor updates, safety layers and data practices that schools cannot inspect. The framework becomes a moral contract without shared governance. Teachers follow rules they did not create; students learn compliance in an environment they cannot contest; and administrators adopt systems they are technically barred from auditing. The enclosure operates through several routes.

This analysis targets a specific configuration of educational AI: closed-weight, cloud-hosted systems that are institutionally integrated through single sign-on, LMS embedding, centralized telemetry and vendor-controlled “policy by update.” In this configuration, schools depend on proprietary change cycles and opaque governance layers, which makes audit rights, change visibility and data reciprocity difficult to secure in practice. The dynamics described here may weaken under alternative arrangements, such as on-prem or locally hosted deployments, open-weight models with institution-controlled modification or public and community-governed consortia that provide shared oversight, transparent update practices and enforceable exit rights. These alternatives do not remove governance risk, but they can reduce the enclosure effects by relocating control over infrastructure, classification thresholds and knowledge gatekeeping closer to educators and communities.

This asymmetry resembles “accountability theater,” a regime of procedural legitimacy in which institutions can demonstrate compliance while lacking audit rights and change control over the systems that shape outcomes. Guidance, in this sense, becomes an instrument of institutional risk management rather than a mechanism of vendor governance. The next section traces that gap through infrastructure, classification, epistemology and labor.

The four lanes of coloniality

I advance a structural claim: educational AI adoption can build and reinforce colonial relations under specific governance conditions through infrastructure, classification, epistemology and labor. Coloniality persists less through intent than through routine arrangements that present themselves as practical and modern: contracts, procurement norms, product defaults and software updates. Quijano (2000) and Mignolo (2011) describe a matrix of power that persists into the present. Big Tech now routes that power through four lanes. The lanes function as a typology of governance mechanisms, with discriminant criteria that keep each lane analytically distinct.

First, infrastructural colonialism forms through cloud contracts, licensing and procurement terms that create long-term dependency. Schools adopt “pilots” that never end, then build workflows, staff roles and student routines around the platform. Exit becomes difficult because identity management, single sign-on, content storage, analytics, assessment and support systems consolidate into one “stack.” Over time, the platform becomes the condition of participation. The institution then pays switching costs in money, time, continuity and credibility.

This dependence produces three governance effects. First, the platform’s update cycle becomes a policy cycle. Rules change on the vendor’s timeline, not on the institution’s calendar. Second, interoperability becomes restrictive. Exports exist, but they often strip context, metadata, permissions history and audit trails, which makes exit formal but not functional. Third, jurisdiction moves outward. When core infrastructure sits in private clouds under vendor terms, schools lose leverage over code audit, data security, redress and transparency. Infrastructure then becomes a hidden lever of corporate governance. This shift reflects the platformization of infrastructure (Plantin et al., 2018), where public utilities are reorganized into private ecosystems that prioritize rent-seeking and data capture over the public mission of the university.

Infrastructure is a pedagogical actor. It teaches students that access to support requires authentication and continuous monitoring. It teaches faculty that instructional help arrives through interfaces designed elsewhere, under rules they cannot revise. The infrastructure becomes a black box system. Over time, the stack normalizes surveillance as a precondition for participation and treats dependence on a small set of firms as normal. This is how governance becomes habit. Students learn that learning is gated, observable and mediated by black box platforms before they learn how to contest those conditions.

Extraction then becomes routine. Student writing, clicks, search queries and revision histories become product telemetry. Vendors can aggregate these traces across institutions, which lets them tune systems and expand market control. Schools do not receive the same returns, because they cannot inspect the full logs, data retained or the downstream reuse of data. This asymmetry turns the performance of ordinary schooling into a resource frontier where public learning produces private advantage.

Classification colonialism operates through the mechanisms of surveillance and prediction. AI systems score, flag and rank students using geographic and linguistic biases that institutions often fail to observe. These tools do more than identify machine text; they enforce a “linguistic enclosure” that treats students writing outside dominant registers as inherently suspicious. By shifting the burden of proof onto the student to disprove an automated accusation of “cheating,” schools create a self-perpetuating cycle: the friction caused by the tool is used to justify deeper institutional reliance on the vendor that provided it. This reflects what Benjamin (2019) calls the New Jim Code, the employment of new technologies that reflect and reproduce existing hierarchies while appearing objective or even “progressive.”

Liang et al. (2023) exposed the depth of this “linguistic enclosure” when they tested seven prominent GPT detectors on 91 TOEFL essays and 88 US eighth-grade essays. Detectors showed near-perfect performance on the US essays, yet misclassified TOEFL essays as AI-generated at an average false-positive rate of 61.22%. In the published version, the authors report that at least one detector flagged 97.8% of TOEFL essays as AI-generated, which turns a technical tool into a disciplinary regime.

In Foucauldian terms, these tools shift assessment toward surveillance and normalization, because they treat deviation from a preferred register as risk that must be documented and corrected (Foucault, 1977). When a system treats predictable language as suspicious, it penalizes constrained expression, translation habits and second-language fluency. In practice, it taxes the Global South and other minoritized writers through higher risk of accusation, higher proof burdens and reduced trust.

Epistemic colonialism concentrates in the safety layers, content availability, response rules and ranking systems set by model developers. These layers operate as gateways in the private governance of inquiry. They decide, in advance, which questions are acceptable, which topics are forbidden, which lines of reasoning receive warnings, and which sources appear and which are treated as authoritative. Interacting with AI systems as a black box, students experience this as an invisible boundary. Educators experience it as guardrails for “responsible use.” Yet the boundary encodes and enforces values they did not choose.

Two mechanisms are key. First, response rules shape the space of possible questions. When a model blocks, softens or deflects certain topics, it edits the horizon of inquiry, often without disclosing the rationale. These controls can reduce harm, yet they also centralize power over access to knowledge. They place decisions about what is discussable in the hands of the system owner and its risk regime, rather than educators and communities. Second, ranking and summarization shape what is valued as knowledge. The model decides which claims surface first, which evidence looks “relevant,” and which perspectives appear marginal or unsafe.

Hila (2025) framed this problem as a mismatch between reliability and understanding. Large language models can approximate a form of externalist justification, yet they do not provide reflective justification because they cannot articulate the basis on which a proposition should be treated as true. When schools outsource reflective work to reliable outputs, they risk eroding reflective standards of justification across the collective, which weakens professional and civic epistemic duties.

The predictable outcome is self-censorship. Students learn to ask what the model will allow rather than the subject they are exploring demands. Teachers learn to design around refusal triggers, which shifts instruction from disciplinary inquiry to platform navigation. Over time, the platform becomes the practical arbiter of “safe curiosity,” which narrows academic freedom through habit rather than decree.

These rules also function as algorithmic paternalism, or even a form of epistemicide (Santos, 2014), the active silencing of certain ways of knowing. The platform decides which knowledge is “safe” to access and which questions require refusal, softening or warning. This decision does not emerge from the classroom, the discipline or the community. It emerges from private risk calculus and generalized assumptions about harm. This creates a shift in authority. Educators lose the capacity to frame difficult knowledge through context and care, while students learn that the boundaries of inquiry are dictated from afar and enforced without the possibility of appeal.

If teachers cannot inspect system prompts, moderation policies, training constraints and update histories, then they cannot teach students how the system helps them to understand “knowledge.” The result is epistemic dependency: learners practice within an information environment whose rules they cannot see, contest or revise.

Finally, labor colonialism manifests as a hidden subsidy to dominant platform firms. Continued profit depends on “quality assurance,” which schools provide without compensation, recognition or governance leverage. Educators and students spend time correcting hallucinations, verifying citations, managing surveillance anxiety, documenting disclosures, rewriting assignments to avoid suspicion triggers and handling misconduct disputes that the tool helps create. This is a prime example of what Ekbia and Nardi (2017) label heteromation, the extraction of economic value from low-cost or free human labor that is mediated by, and hidden behind, automated systems.

This falls beyond “extra work” that is a normal part of using these tools, becoming a form of wealth transfer. Teacher time is public capital. It is funded to support learning, not to stabilize commercial systems that remain opaque to the people forced to rely on them. When educators perform quality assurance, redesign assessment around platform limits and manage authorship disputes triggered by automated suspicion, they convert public labor into private product quality assurance and stress testing. The vendor captures the gains through adoption, renewals and market legitimacy, while schools absorb the costs as burnout, diverted attention and reduced capacity for educating.

Table 1. Visualizing the four lanes of coloniality

Note. Created by author.

Consider this vignette: A student submits a draft. An AI tool detector flags it. The instructor pauses grading, meets the student, reviews version history and writes a memo for an integrity process. A chair reviews the memo, a committee schedules a hearing and the student rewrites under surveillance. The vendor does not attend, does not disclose the thresholds that triggered the inquiry and does not compensate the hours spent stabilizing trust. The institution funds the labor, while the vendor retains the authority that helped produce the dispute and may retain the drafts for its own training.

The illusion of “using” AI and the limits of “ethics”

Generative AI can deliver real instrumental benefits in educational settings, including accessibility supports, such as language assistance and alternative formats, drafting support and rapid formative feedback. These benefits strengthen the case for governance, because reliance without audit rights increases harm when failures occur. When institutions cannot inspect system behavior, contest classification thresholds or track changes in safety and ranking logic, the same tools that support learning can also amplify inequity, trigger unjust suspicion and narrow inquiry without remedy.

Schools often describe AI as a tool that educators “use.” The language implies agency: the educator acts to further their own goals and the tool assists. Yet platforms steer action through defaults that shape attention and judgment. A dashboard summarizes progress around narrow indicators. A safety layer defines what may be permissible inquiry. A recommendation system determines relevance. These mechanisms govern through interface and default settings, which shifts authority from deliberative policy processes to vendor-controlled design choices. In essence, they perform the pedagogy rather than assist it.

This is how colonization becomes normalized in practice. Long before a teacher writes a prompt, the platform has already set the boundaries of the safe, the visible and the credible. Educators then manage classroom interactions as if those boundaries were acceptable. Students learn what questions lead to refusal by the system, what language avoids suspicion and what forms of argument the system rewards. The platform reorganizes instruction around its own rule set.

This pattern persists because decisions that shape classrooms occur upstream. Procurement clauses that waive audit rights, default safety policies that filter topics without educator input, telemetry that records student behavior and dashboards that define progress through vendor metrics all function as an invisible curriculum. Where power sits, in settings, thresholds and update processes, is where governance has been ceded.

Yet educators and students are told to use AI “ethically.” Institutional AI ethics frameworks often function as a ruse. They place moral and professional responsibility on teachers and students while leaving vendors in unilateral control of model behavior, data flows and system change. The framework then becomes a display of compliance theater. It rewards vigilance, documentation and disclosure, yet it cannot govern the systems that generate the risk. The result is a loss of agency: educators and students have fewer ways to act, fewer reasons to trust their judgment and fewer channels for challenging the system.

This displacement of accountability follows a colonial logic: institutions discipline users at the edge while platform power remains upstream and unaccountable. A response grounded in ethics alone cannot correct that asymmetry. The central task is democratic governance of systems that set the conditions of inquiry, data use and change. This requires a shift from adaptation to authority, so schools can set, review and revise the conditions under which AI operates in learning.

Reclaiming agency through sovereignty

Agency is the capacity to shape the conditions of learning. Self-efficacy is the learned belief, grounded in competence and feedback, that one’s actions can produce results (Bandura, 1997). Platforms weaken both when they narrow options, hide system behavior and turn learning into compliance with opaque rules. Schools can rebuild both when they make governance visible and participatory and when they teach students how to interrogate systems rather than submit to them.

Liberation from the four lanes of coloniality requires an institutional shift from readiness to sovereignty. While readiness merely measures the scale of access, sovereignty measures the depth of control and review. As Smith (2021) argues, decolonizing methodologies require the reclamation of control over how one’s community is represented and governed. In education, this means refusing the role of the “passive user” and asserting the right to “design the conditions of our own inquiry” (Simpson, 2014).

Policy discourse often measures devices, bandwidth and platform adoption. Yet a system can look “ready” while remaining governed from outside. When schools rely on external AI infrastructure that they cannot inspect, question or revise, they import governance, functionality and responsibility. They adopt the platform’s update cycle, its goals and its metrics of performance. Readiness then becomes the language through which dependence presents itself as progress.

Schools often treat sovereignty as a technical matter and hand it to vendors and administrative compliance staff, but it should start in the classroom. It needs to begin with setting the limits of inquiry, and who can change them. In Table 2, I share a simple diagnostic that any school should be able to complete. The answers help show to what extent effective edtech governance exists and is intended to serve as a launch pad for further investigations.

Table 2. A sovereignty diagnostic for collective educational technology governance

Answer each question: Yes, Partly, or No.

Note. Created by author.

A minimum sovereignty agenda begins with procurement. Institutions and systems should refuse click-wrap licenses for high-stakes use and require clauses that guarantee audit rights, review of change logs and minimal use of data collected from students and the institution. The black box of the system needs to be made transparent. Model documentation should disclose intended use, known failure modes and rule behaviors, so educators can teach within the system’s limits rather than guess at them (Mitchell et al., 2019).

Further, sovereignty requires a redistribution of responsibility aimed at increasing educator and student agency. If institutions continue to insist on using tools that detect suspicious behavior, they should establish independent evaluation and appeal rights for students flagged by automated scoring or detection, with remedies that are documented and enforceable. Pedagogical repair labor should appear as funded work in budgets and balance sheets, not as invisible surplus effort. Adoption boards should include educator and student voting roles with authority to pause or terminate tools that erode academic freedom, accessibility or equitable participation.

Recent work on AI governance in education identifies a related institutional problem: use policies alone cannot govern adoption. Effective governance requires procurement oversight, auditability, data protection, change visibility and accountability across the system lifecycle. Studies of higher education policy frameworks show that institutions must move from ad hoc classroom guidance toward coördinated governance, policy review, risk assessment and stakeholder oversight (Tong et al., 2025; Wu et al., 2024). Broader AI governance literature also emphasizes that accountability must be assigned across the AI lifecycle, including what is governed, who is responsible, when governance occurs and how oversight is enacted (Batool et al., 2023). In procurement, these concerns become concrete through questions of data use, ownership, vendor transparency, black-box behavior, cost and institutional risk (Grajek et al., 2025). These issues carry economic consequences. Vendor lock-in raises exit costs, opaque updates shift operational risk onto institutions and unpaid repair labor transfers public educational value into private platform stabilization. A sovereignty agenda therefore treats governance as part of the cost of adoption, not as an afterthought once tools have entered classrooms.

Toward a collective praxis

Confronting educational AI should not be viewed as a technical problem to solve with better prompts or “responsible use” modules. It is an exercise of collective praxis, critical reflection joined to institutional action (Freire, 2000). Asserting sovereignty is not a rejection of technology, but a decolonizing methodology (Smith, 2021) that centers the community’s right to define the boundaries of its own intellectual life. This constitutes a governance struggle over the digital conditions of learning. The enclosure of the classroom does not follow from technology itself. It follows from a structural logic of extraction that normalizes dependence and then asks educators and students to manage the consequences.

Education now faces a choice between steering its intellectual future and being steered deeper into a colonial matrix of platform power. Liberation requires more than readiness. It requires the capacity to refuse defaults, contest opaque rules and rebuild public control over infrastructure, data and inquiry. Through collective praxis, critical reflection joined to institutional action, schools can interrupt the four lanes of coloniality and restore agency and self-efficacy as educational outcomes. Regaining educational sovereignty is the condition for keeping the university a public good rather than a resource frontier for private capital.

Author’s Statement of Algorithmic Interlocution and Sovereignty

In alignment with the minimum sovereignty agenda proposed in this work, the author provides the following disclosure regarding the participation of generative AI (ChatGPT 5.5 and Gemini 3 Flash) in the production of this manuscript:

• Role of the interlocutor. AI served as an algorithmic interlocutor during drafting and revision. It supported stress-testing of the four-lanes framework, generated alternative phrasings to improve clarity and assisted with targeted copyediting.
• Labor of repair. The author performed sustained repair labor throughout. Machine-generated suggestions that drifted toward institutional isomorphism, generic moralizing or unauditable certainty were rejected. The conceptual architecture of the argument, including the four-lanes framework, the synthesis of colonial theory and the sovereignty diagnostic, originated with the human author.
• Auditability and agency. The author retained jurisdiction over all upstream decisions, including the manuscript’s claims, definitions and argumentative structure. No passage was accepted as black-box output. Every suggestion was reviewed for conceptual accuracy, rhetorical alignment and avoidance of platform-shaped bias that could narrow inquiry.
• Telemetry and extraction. The author recognizes that the interaction history generated during drafting likely produced telemetry for the platform. This constitutes a practical example of the extraction dynamics discussed in the manuscript, in which public intellectual labor can be repurposed for the benefit of private systems.
• Final authority. The author retains full sovereignty over the final text and accepts sole responsibility for its content, interpretations and conclusions.

References

• Andrejevic, M. (2007). Surveillance in the digital enclosure. The Communication Review, 10(4), 295–317. https://doi.org/10.1080/10714420701715365
• Bandura, A. (1997). Self-Efficacy: The Exercise of Control. W. H. Freeman.
• Batool, A., Zowghi, D., & Bano, M. (2023). Responsible AI Governance: A Systematic Literature Review. arXiv. https://doi.org/10.48550/arXiv.2401.10896
• Benjamin, R. (2019). Race after Technology: Abolitionist Tools for the New Jim Code. Polity.
• DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160. https://doi.org/10.2307/2095101
• Ekbia, H. R., & Nardi, B. A. (2017). Heteromation and Other Stories of Computing and Capitalism. MIT Press. https://doi.org/10.7551/mitpress/10767.001.0001
• Foucault, M. (1977). Discipline and Punish: The Birth of the Prison. Vintage.
• Freire, P. (2000). Pedagogy of the Oppressed (30th anniversary ed.). Continuum. Original work published 1970.
• Grajek, S., Pelletier, K., & Freeman, A. (2025). AI procurement in higher education: Benefits and risks of emerging tools. EDUCAUSE Review. https://er.educause.edu/articles/2025/3/ai-procurement-in-higher-education-benefits-and-risks-of-emerging-tools
• Hila, A. (2025). The epistemological consequences of large language models: Rethinking collective intelligence and institutional knowledge. AI & Society, 41(1), 1–19. https://doi.org/10.1007/s00146-025-02426-3
• Liang, W., Yuksekgonul, M., Mao, Y., Wu, E., & Zou, J. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7), Article 100779. https://doi.org/10.1016/j.patter.2023.100779
• Mignolo, W. D. (2011). The Darker Side of Western Modernity: Global Futures, Decolonial Options . Duke University Press.
• Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman, L., Hutchinson, B., Spitzer, E., Raji, I. D., & Gebru, T. (2019). Model cards for model reporting. In Proceedings of the Conference on Fairness, Accountability, and Transparency (pp. 220–229). Association for Computing Machinery. https://doi.org/10.1145/3287560
• Plantin, J.-C., Lagoze, C., Edwards, P. N., & Sandvig, C. (2018). Infrastructure studies meet platform studies in the age of Google and Facebook. New Media & Society, 20(1), 293–310. https://doi.org/10.1177/1461444816661553
• Quijano, A. (2000). Coloniality of power, Eurocentrism, and Latin America. Nepantla: Views from South, 1(3), 533–580.
• Santos, B. de S. (2014). Epistemologies of the South: Justice against Epistemicide. Routledge.
• Simpson, L. B. (2014). Land as pedagogy: Nishnaabeg intelligence and rebellious transformation. Decolonization: Indigeneity, Education & Society, 3(3), 1–25.
• Smith, L. T. (2021). Decolonizing Methodologies: Research and Indigenous Peoples. Bloomsbury Publishing. https://doi.org/10.5040/9781350225282
• Tong, A., Zainol, Z., Chong, T. S., & Renganathan, K. (2025). AI governance on young consumers in higher education: A content analysis of policies for generative AI. Young Consumers, 26(5), 865–885. https://doi.org/10.1108/YC-10-2024-2303
• Wu, C., Zhang, H., & Carroll, J. M. (2024). AI governance in higher education: Case studies of guidance at Big Ten universities. Future Internet, 16(10), Article 354. https://doi.org/10.3390/fi16100354
• Zuboff, S. (2019). The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power . PublicAffairs.

Further reading

• Couldry, N., & Mejias, U. A. (2019). The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism . Stanford University Press.
• Haraway, D. (2013). Situated knowledges: The science question in feminism and the privilege of partial perspective. In Women, Science, and Technology (pp. 455–472). Routledge.
• Harding, S. (1991). Whose Science? Whose Knowledge? Thinking from Women’s Lives. Cornell University Press.
• hooks, b. (2014). Teaching to Transgress: Education as the Practice of Freedom. Routledge. https://doi.org/10.4324/9780203700280
• Lyotard, J.-F. (1984). The Postmodern Condition: A Report on Knowledge. G. Bennington & B. Massumi, Trans. University of Minnesota Press.
• Nkrumah, K. (1965). Neo-Colonialism: The Last Stage of Imperialism. International Publishers.
• Noble, S. U. (2018). Algorithms of Oppression: How Search Engines Reinforce Racism. NYU Press. https://doi.org/10.18574/nyu/9781479833641.001.0001
• Selwyn, N. (2021). Education and Technology: Key Issues and Debates. Bloomsbury Publishing.
• Zimmerman, B. J. (2000). Attaining self-regulation: A social cognitive perspective. In M. Boekaerts, P. R. Pintrich, & M. Zeidner (Eds.), Handbook of Self-Regulation (pp. 13–39). Academic Press. https://doi.org/10.1016/B978-012109890-2/50031-7