Technical Brief · May 2026
A Technical Brief on Community-Owned, Self-Funding Digital Infrastructure for School Modernization in the Rural Global South
The effort to connect every school to the internet is among the most ambitious infrastructure undertakings of the current era. Connectivity, however, is a starting point, not a finish line. A school that is connected to the internet but dependent on proprietary hardware, locked into a single vendor's ecosystem, or generating data that is harvested without consent or compensation has not been modernized. It has been plugged into an extractive system by different means.
This brief addresses what comes after the connection is live: who owns the infrastructure, how it is funded, and what architectural standards must be in place to ensure that no harm reaches the child. The answers draw on existing, deployed technologies that have been built and tested in the regions where they would be applied. None require a single vendor. None require external debt. None require permission from any authority that would withhold it.
The frameworks, specifications and reference implementations described here are placed in the public domain and may be used, adapted and deployed by any government, multilateral institution or community organization without attribution, payment or permission.
The effort to connect every school to the internet is among the most ambitious infrastructure undertakings of the current era. Initiatives such as GIGA, a partnership between UNICEF and the International Telecommunication Union, have made measurable progress in mapping school locations, assessing connectivity gaps, and negotiating bulk connectivity agreements with telecommunications providers. This work is necessary. It deserves acknowledgement and support.
Connectivity, however, is a starting point, not a finish line. A school that is connected to the internet but dependent on proprietary hardware, locked into a single vendor's ecosystem, or generating data that is harvested without consent or compensation has not been modernized. It has been plugged into an extractive system by different means.
This technical brief addresses three questions that arise once the connection is live:
The answers offered here are not theoretical. They draw on existing, deployed technologies: community-owned mesh networks, anonymized data licensing frameworks, and stateless-by-design protocols that have been built and tested in the regions where they would be applied. None require a single vendor. None require external debt. None require permission from any authority that would withhold it.
The document is structured around four principles: self-funding through sovereign data assets, beginning with a single strategically chosen data stream; architectural enforcement of safety, privacy and agnosticism; community ownership as a minimum standard; and reciprocity mechanisms that allow communities, counties and regions to federate their infrastructure for mutual support, reducing the burden on national governments and eliminating extraction layers between the resource and the child.
This brief is published as a public record and a standing offer. The solutions described exist. They have been built. They are available to any partner, multilateral, governmental or private, prepared to implement them in good faith.
Infrastructure built for schools should be owned by the communities those schools serve. This is not a concession to be negotiated. It is the minimum condition for any intervention that purports to serve the interests of children.
Community ownership is distinct from community consultation, community benefit or community participation. Consultation asks for input. Benefit promises outcomes. Participation invites involvement. Ownership confers control. A community that owns its school's digital infrastructure can determine how it is used, who may access it, under what terms data may be shared, and whether a vendor's contract is renewed. No external entity can withhold connectivity as leverage. No third party can extract rent from a resource the community itself maintains.
This principle is not new. Prior to the disruptions of the colonial period, communities across what is now termed the Global South maintained sophisticated systems of education, knowledge transfer and resource governance. These systems were self-sustaining, adapted to their contexts, and answerable to the people they served. The present condition of educational infrastructure deficit is not a starting point. It is an interruption. Community ownership of digital infrastructure is not an innovation. It is a restoration.
Community ownership of connectivity infrastructure is technically achievable with existing, off-the-shelf hardware. One demonstrated model is the nexus mesh: a peer-to-peer wireless network connecting multiple schools or community anchor points without dependence on any single telecommunications provider.
The nexus mesh draws on proven implementations, including the wireless mesh networks deployed in Kenya and the Democratic Republic of the Congo through initiatives such as AfriNet. These networks demonstrated that communities can build and maintain their own connectivity infrastructure when provided with appropriate hardware, basic training and the legal standing to operate.
A nexus mesh consists of solar-powered edge nodes located at schools or community centres. Each node creates a local network that caches educational content, runs local applications and handles authentication without requiring constant internet connectivity. Synchronization with the broader internet occurs through whatever transport is available: a satellite link, a fibre connection at a central point, or, in the most resource-constrained settings, a portable storage device transported by motorcycle between nodes. This asynchronous-by-design architecture means that a school does not lose functionality when the connection drops. Learning continues. Records remain local. The mesh is resilient by structure, not by contingency.
Each node is owned by the school or community that hosts it. The mesh is the sum of these owned pieces cooperating through standard, open protocols. No central provider controls access. No entity can impose fees, throttle bandwidth or monetize usage data. The infrastructure is the community's own, in the same way a community well or a shared solar array is its own.
For community-owned infrastructure to function at scale, national governments must establish two enabling conditions.
First, communities must have legal standing to own and operate connectivity infrastructure. This may require amendments to telecommunications regulations that currently reserve spectrum access and infrastructure operation to licensed commercial entities. Community spectrum licensing, particularly in television white space frequencies, which are well-suited to the distances and terrain of rural areas, provides a regulatory model that several nations have already adopted or piloted.
Second, schools must be recognized as valid sites for infrastructure ownership, distinct from household or commercial premises. A school is a community anchor institution. Its connectivity should be governed accordingly.
Neither condition requires significant public expenditure. Both require political will. The return on that will is a connectivity model that costs the national government less over time, not more, because the burden of maintenance and operation is carried by the communities that benefit.
When infrastructure is owned by a vendor, every decision about that infrastructure passes through a commercial filter. Will the vendor support older hardware, or will it require proprietary upgrades? Will the vendor allow interoperability with other systems, or will it design its products to preclude it? Will the vendor continue to serve a remote community when the contract's margins tighten, or will it withdraw?
These are not hypothetical questions. They describe the documented behaviour of technology vendors in education markets globally. The only structural safeguard against these outcomes is ownership. A community that owns its infrastructure cannot have it withdrawn. A community that maintains its infrastructure cannot have access priced beyond its means. A community that governs its infrastructure cannot have its children's data harvested by a distant entity that has never set foot in its school.
Any infrastructure deployed under this principle should meet the following minimum specifications:
There is an asset already generated daily by millions of children attending school, currently given away without compensation. That asset is language data.
Infrastructure requires investment. The conventional mechanisms for financing digital infrastructure in the Global South, sovereign debt, foreign aid and vendor-financed procurement, each carry structural consequences. Debt creates obligations that outlast the infrastructure it funds. Aid is subject to the priorities of the funder, not the recipient. Vendor financing embeds extraction into the operational model from the first day of deployment.
There is an alternative. It does not require new taxes, external borrowing or philanthropic generosity. It requires recognizing an asset that already exists, that is generated daily by millions of children attending school, and that is currently given away without compensation. That asset is language data.
Global artificial intelligence systems are trained predominantly on English-language data. The world's languages, numbering more than 7,000, the majority spoken in the Global South, are dramatically underrepresented in training corpora. This is not a curiosity. It is a market inefficiency of enormous scale.
Technology companies, research institutions and governments investing in AI development require diverse, high-quality language data to build systems that function beyond a narrow set of dominant languages. They require spoken language, written language, signed language. They require it across dialects, age groups and contexts. They require it from authentic speakers engaged in authentic communication: precisely the kind of communication that occurs naturally in classrooms.
This data is non-rivalrous. Licensing it does not deplete it. It is renewable, generated continuously through the act of teaching and learning. And it is scarce in the sense that matters to markets: it cannot be synthesized elsewhere. Only the communities that speak, write and sign these languages can produce authentic, representative data.
At present, this data is extracted without consent, without compensation and without benefit to the communities that produce it. Platforms that collect user-generated content harvest language data as a by-product of their terms of service. The value flows out. Nothing flows back.
A sovereign Educational Data Trust inverts this flow. The mechanism is straightforward in concept, though it requires careful legal and technical implementation:
The trust does not sell children's data. It licenses anonymized, aggregated data that cannot be traced back to any individual student. It does not permit the data to be used for surveillance, profiling or commercial targeting. It embeds these prohibitions in the licensing agreement with contractual penalties for violation. The trust is not a vendor. It is a steward of a sovereign asset, accountable to the public whose asset it manages.
A government need not commodify every aspect of education to fund it. A single, strategically chosen data stream can generate revenue sufficient to modernize a national school system, provided three conditions are met.
First, the licensing framework must capture fair market value. This requires that nations negotiate collectively where possible, rather than competing against each other to underprice their data. Regional data trusts covering multiple nations with shared language families offer a mechanism for collective bargaining.
Second, revenue must be constitutionally or legislatively earmarked for education. Without a legal firewall, licensing revenue will be absorbed into general government expenditure, and the infrastructure it was intended to fund will go unfunded. Earmarking is not a technical detail. It is the mechanism that makes the model self-sustaining.
Third, the infrastructure that generates and transmits the data must be community-owned, as described in Principle One. If the infrastructure is vendor-owned, the vendor will extract its own rents, and the revenue that reaches schools will be a fraction of what the data is worth. Community ownership closes the loop. The same infrastructure that delivers curriculum also generates the asset that funds it.
The Educational Data Trust model is not without precedent. Elements of this approach are already in operation or under active development:
None of these precedents is a complete Educational Data Trust. Each provides a component of the model. The task is to assemble the components and adapt them to the educational context.
The Educational Data Trust is not a mechanism for extracting value from children. It is the opposite: a mechanism for preventing extraction by establishing a lawful, transparent, community-governed alternative.
It is not a substitute for public investment in education. Governments retain their obligation to fund schools. The trust supplements public expenditure with revenue that would otherwise not exist, earmarked for infrastructure that would otherwise not be built.
It is not a gift from the technology sector. It is a commercial transaction in which a willing buyer licenses an asset from a willing seller on terms set by the seller. The fact that the seller has historically given the asset away for free does not oblige it to continue doing so.
A policy promise is revocable. A technical control, properly implemented, is not. Safety, privacy and sovereignty must be embedded directly into the technical fabric of infrastructure, not added as an afterthought.
The Global Dialogue on AI Governance has called for a paradigm shift from "governance by vibes" to architectural enforcement: the embedding of safety, privacy and sovereignty directly into the technical fabric of infrastructure. This principle applies with particular force to schools. Children are not a test market. Classrooms are not a proving ground for emerging technologies. The standards that govern digital tools in education must be non-negotiable, verifiable and built into the architecture of the tools themselves.
The following six standards constitute the minimum architectural requirements for any digital tool deployed in a school, regardless of vendor, regardless of jurisdiction, regardless of funding source. They are drawn from existing, deployed implementations. They are achievable today with off-the-shelf hardware and open-source or freely licensable software. A vendor that cannot meet them is not qualified to serve children.
Any digital tool used in a school must process student data locally, with zero persistence of personally identifiable information in any cloud environment, unless explicit, revocable consent has been obtained from the data subject or their legal guardian, in accordance with national law and the Convention on the Rights of the Child.
In practice, this means:
Stateless-by-design architecture moves privacy from a legal debate to a technical certainty. It protects children in settings where connectivity is intermittent, where data protection laws are nascent or unenforced, and where the balance of power between a vendor and a family is fundamentally unequal.
No biometric data, including facial recognition, fingerprints, iris scans, voice prints or gait analysis, shall be collected, stored or processed from any person under the age of 18 in an educational setting. This prohibition is absolute. It does not admit exceptions for safety, convenience or administrative efficiency.
A child's body is not a password. Biometric systems, once deployed in schools, cannot be meaningfully consented to by minors, cannot be effectively opted out of without stigma or exclusion, and create databases of irrevocable identifiers that follow children into adulthood. The security benefits claimed by vendors of these systems are consistently overstated. The risks, including data breach, function creep and the normalization of surveillance as a condition of education, are structural and irreversible.
This prohibition applies regardless of whether the biometric processing occurs locally or in the cloud, regardless of the vendor's assurances, and regardless of whether the system has been deployed in other jurisdictions. It is a bright line. Any vendor that proposes biometric systems in schools is disqualified.
Digital tools must function with any school computer and any compatible peripheral device, not only the hardware sold or approved by the tool's vendor. Software must be fully white-label: it does not display the vendor's brand, logo or product names inside the school environment, unless the school expressly chooses otherwise.
Vendor lock-in is not an accident. It is a business model. Proprietary hardware requirements, closed file formats and brand propagation inside schools are designed to make switching costs prohibitive and to extend the vendor's commercial relationship indefinitely. Agnosticism as an architectural standard prevents lock-in before it begins. A school that deploys a digital tool today must be able to replace it tomorrow without losing access to its data, its hardware or its instructional continuity.
This standard applies to every vendor without exception, including any entity associated with the authors of this brief. Software agnosticism is not optional. The tool must serve the school. The school must not advertise the tool.
All student data must be processed locally on school-owned or community-owned devices. No student data may be used for training vendor AI models, for commercial profiling, for targeted advertising or for any purpose other than the direct educational function of the tool. No student intellectual property, including creative work, written assignments or project outputs, may be licensed, sold or incorporated into a vendor's training corpus without the explicit, compensated consent of the student or their legal guardian.
This standard addresses a documented pattern: education technology vendors that offer "free" tools in exchange for data rights buried in terms of service. The tools are not free. The price is the data. The transaction is obscured. Local processing as an architectural default makes extraction technically impossible, regardless of what the terms of service claim or omit.
Any software deployed in a school must be open-source or available for independent security audit by a qualified third party selected by the procuring authority. Proprietary black boxes have no place in educational infrastructure.
This is not an ideological position. It is a practical one. Schools cannot verify what proprietary software does with student data. They cannot confirm that it meets the standards described above. They cannot detect whether it has been altered, whether it contains vulnerabilities or whether it is transmitting data to parties unknown. Auditability is the mechanism that makes all other standards enforceable. Without it, standards are trust. With it, standards are verification.
No digital tool shall be activated in a classroom until educators, administrators and, where age-appropriate, students have received role-specific, sustained training on its use, its limitations and the rights and protections associated with it. Training is not an afterthought. It is not a one-time orientation session. It is a prerequisite to deployment.
A tool placed in a classroom without trained adults is not a learning intervention. It is an experiment. The adults responsible for children must understand what the tool does, what it does not do, what data it processes, where that data resides and what to do when something goes wrong. Training must be ongoing, updated as tools evolve, and delivered in the language of instruction.
Each of the six standards described above has been implemented in working systems. Stateless-by-design protocols are in operation in secure communications infrastructure. Local processing of AI models runs on devices powered by five watts of electricity. Open-source educational platforms serve millions of learners without extracting their data. Training before deployment is standard practice in every profession that takes harm prevention seriously: medicine, aviation, engineering. Education should be no different.
Communities federate their infrastructure for mutual support under agreed protocols: horizontal exchanges between equal parties, with no extraction layer, no vendor intermediation and no dependency created that cannot be unwound.
A community that owns its infrastructure and funds it through sovereign data assets is already more resilient than one dependent on vendors, debt or aid. But no community is an island. Schools exist within counties, counties within regions, regions within nations. The connectivity, compute power and technical expertise available to one community will differ from that available to its neighbour. Mechanisms that allow communities to share what they have, voluntarily, transparently and without creating new dependencies, extend the benefits of the model to those who would otherwise go without.
Reciprocity, in this context, means that communities federate their infrastructure for mutual support under agreed protocols. A community with surplus connectivity shares bandwidth with a community that lacks it. A school with underutilized compute capacity on its edge node makes that capacity available to a neighbouring school during off-peak hours. A region that has trained technical personnel lends that expertise to a neighbouring region while it builds its own capacity.
These arrangements are not charity. They are not aid. They are horizontal exchanges between equal parties, governed by transparent agreements that either party can exit without penalty. If one community has more than it needs and another has less, the surplus should flow to the need; not as a gift that creates obligation, but as a shared investment in a system that benefits all participants.
A reciprocity agreement between communities requires three elements.
Technical interoperability. Community-owned infrastructure built on open protocols is, by design, interoperable with other infrastructure built on the same protocols. A nexus mesh node in one school can communicate with a node in a neighbouring school without proprietary translation layers, without vendor intermediation and without fees. The technical standards described in Principle Three, specifically hardware agnosticism, open or auditable code and stateless-by-design architecture, are the same standards that make federation possible.
Transparent governance. Reciprocity arrangements must be governed by written agreements that specify what is shared, under what conditions, for what duration, and with what process for amendment or termination. The agreements must be public documents, accessible to the communities they affect, and subject to periodic review. Obscurity creates suspicion. Transparency builds trust.
Mutual benefit. A reciprocity arrangement that consistently benefits one party at the expense of the other is not reciprocity. It is extraction by another name. Agreements should be structured so that over time, each participating community both contributes and receives. The contribution need not be identical; one community may contribute bandwidth while another contributes maintenance labour. But the balance must be recognizably mutual.
When communities can meet their own needs through mutual support, the burden on higher levels of government diminishes. A national education ministry that does not need to fund, procure and maintain connectivity for every school can direct its resources to functions that only a central authority can perform: setting curriculum standards, accrediting teachers, ensuring equity across regions, and maintaining the legal and regulatory framework within which community-owned infrastructure operates.
This is subsidiarity in practice: the principle that decisions and resources should reside at the most local level capable of managing them effectively. A community can manage its own connectivity. A county can coordinate reciprocity between its communities. A region can maintain the edge infrastructure that requires scale. The national government can set the standards and enforce the protections. Each level does what it is best positioned to do, and no level is burdened with responsibilities that properly belong elsewhere.
Reciprocity is not a substitute for public investment. Governments retain the obligation to fund education and to ensure that every child, regardless of location, has access to a school that meets minimum standards. Reciprocity reduces the cost of meeting that obligation. It does not eliminate the obligation.
Reciprocity is not a mechanism for offloading government responsibility onto communities that cannot bear it. A community that lacks the resources to build its own infrastructure cannot be told to rely on the generosity of its neighbour and left to manage as best it can. The national government retains a duty to address structural inequities between regions. Reciprocity handles the margins. Equity requires the centre.
Reciprocity is not a vendor relationship by other means. A community that shares bandwidth with a neighbour is not selling a service. It is participating in a mutual aid arrangement that either party can exit. There is no contract that locks in terms. There is no extraction of profit. There is no dependency created that cannot be unwound.
A single school with a nexus mesh node is a proof of concept. Two schools sharing bandwidth are a pilot. Twenty schools across a county, federated by reciprocity agreements, with a shared Educational Data Trust generating revenue that funds maintenance and expansion: that is a system.
The trajectory from proof of concept to system does not require central planning. It requires standards, legal frameworks and a few working examples that others can adapt. Communities learn from communities. A reciprocity agreement that works in one county is copied in the next. A governance structure that proves effective in one region is adopted by its neighbour. The role of national government is to enable this diffusion, not to direct it; to set the rules, enforce the protections, and then step back.
This addendum provides practical guidance for governments, multilateral institutions and other duty-bearers responsible for procuring digital infrastructure for schools. It is not exhaustive. It is a baseline. Any procurement that does not meet these specifications should be reconsidered.
| Requirement | Specification |
|---|---|
| Power Independence | Infrastructure must operate on solar power with battery storage sufficient for continuous operation during school hours and a minimum of four hours beyond. Grid dependency is a failure point. Solar is a requirement in any setting where power supply is unreliable, which is to say the majority of rural settings in the Global South. |
| Asynchronous Operation | Infrastructure must provide full local functionality, including content delivery, learning management, assessment and data collection, without continuous internet connectivity. Synchronization must occur through standard protocols that accommodate intermittent connections, including peer-to-peer transfer via portable storage media. |
| Open Networking Standards | All wireless communication must use open, non-proprietary protocols (e.g., IEEE 802.11 standards for local networking). Proprietary networking protocols that require vendor-specific hardware to extend or maintain the network are not acceptable. |
| Auditable Firmware | All firmware must be open-source or available for independent security audit. The procuring authority retains the right to commission an audit at any time, at its own discretion, by an auditor of its own selection. |
| Modular Hardware | Components must be replaceable with off-the-shelf equivalents. No single component may be designed such that its failure requires replacement of the entire unit or procurement from a single supplier. |
| Physical Durability | Hardware must be rated for the environmental conditions of the deployment site, including temperature, humidity and dust exposure. A vendor that ships hardware designed for air-conditioned server rooms to schools without reliable electricity has not solved a problem. It has created one. |
The following questions should be put to any vendor or implementing partner. A vendor that hesitates, deflects or answers "no" to any of these questions should not be considered qualified. The procuring authority is not obligated to accept an inferior system because a vendor has an appealing presentation.
In addition to the frameworks described in this brief, the author has developed a backend infrastructure solution for identifying and disrupting child trafficking and labour exploitation networks. The solution was submitted to relevant authorities in 2023 and was declined on procedural grounds unrelated to its technical merit, specifically the nationality of its author.
The solution remains available, at no cost, to any legitimate implementing body, public, multilateral or private, prepared to deploy it in good faith. It is mentioned here not as a grievance but as a standing offer. The protection of children from exploitation is too urgent to be delayed by administrative theatre. The door remains open.
The following projects and initiatives demonstrate components of the architecture described in this brief. They are cited as evidence that the specifications above are technically achievable, not as endorsements of any particular implementer.
The solutions described in this brief exist. They have been built, tested and in several cases deployed at scale. They do not require new inventions, new materials or new scientific breakthroughs. They require only the willingness to implement what is already possible.
This brief does not ask for funding. It does not ask for debt relief. It does not ask for preferential treatment in procurement. It does not ask for a seat at any table that is not willingly offered.
What it asks is simpler: that the institutions and governments entrusted with the welfare of children consider the evidence, evaluate the specifications and act on what they find. If the model described here is sound, and it is testable by any party with the technical competence to evaluate it, then the question is not whether it can be done. The question is whether it will be.
The frameworks, specifications and reference implementations described in this brief are placed in the public domain. They may be used, adapted and deployed by any government, multilateral institution or community organization without attribution, without payment and without permission.
In addition, the author offers the following:
None of these offers is contingent on employment, contract or institutional affiliation. They are made in good faith, to any party that shares the objective of modernizing educational infrastructure without extraction, without dependency and without harm to children.
The nations of the Global South are not poor because they lack resources. They are poor because the mechanisms by which resources are converted into prosperity have, for centuries, been controlled by forces external to them. Colonization extracted physical wealth. Structural adjustment extracted fiscal sovereignty. Data extraction, in its current form, extracts the raw material of the next economy: language, culture, knowledge, without consent, without compensation and without leaving behind the infrastructure that would make further extraction unnecessary.
This is not a natural state of affairs. It is a produced condition. And what has been produced can be unmade and remade differently.
Every day, millions of children walk into schools across the Global South. They speak. They write. They sign. They learn. They generate, simply by being present, an asset of extraordinary value. That asset currently flows outward, enriching those who already have more than enough.
It could, instead, flow to the children who produce it: building the schools they learn in, powering the devices they use, training the teachers who guide them, and funding the next school in the next community and the next after that.
The mechanism exists. The architecture is specified. The offer is on the table.
The only remaining question is who will take it up.
This brief was prepared as a contribution to the ongoing discourse on sustainable educational infrastructure in the Global South. It draws on the author's experience in digital public infrastructure, the modernization frameworks previously advocated for schools in the United States, and the position submitted to the Global Dialogue on AI Governance.
The author holds degrees in communications and has worked for several years on initiatives aligned with the Sustainable Development Goals. This brief is published as a public resource. It is not commissioned, not remunerated, and not submitted on behalf of any organization. The offer it describes is made in a personal capacity and in good faith.
Contact: verity@hithertoai.org · hithertoai.org