THE SCENARIO
A mid-sized national health ministry in the Global South adopted a cloud-hosted AI diagnostic platform from a major US technology provider. The platform promised improved patient outcomes through machine-learning image analysis. Three years into operational use, the ministry discovered that its patient data — including records of indigenous communities — were being used to train the provider’s commercial models without explicit consent or compensation. When the ministry attempted to renegotiate data governance terms, it discovered that the contract included a mandatory foreign-law data access clause that prevented data isolation without contractual penalty.
To sever the relationship and deploy an alternative would require retraining the diagnostic model on locally sourced medical imaging data — a process estimated at eighteen months and twelve million dollars, an investment the ministry could not justify within its annual budget cycle. The platform had become infrastructure not through design but through a sequence of individually reasonable decisions that collectively removed viable alternatives.
A further development illustrated the full extent of the dependency: when the provider updated its pricing model, the cost of the platform increased by three hundred and forty percent. The institution faced a binary choice: absorb the increase and divert resources from other programmes, or accept the cost of rebuilding analytical capacity that had been allowed to atrophy. Neither option reflected genuine sovereign choice. Both were products of a dependency architecture that had been constructed without the institution’s full awareness.
The Question
What does technology dependency actually cost — and who bears that cost when the dependency becomes structural?
This article examines the mechanisms by which AI platforms convert adoption into dependency, assesses the costs that dependency imposes on sovereign entities, and offers a structured framework for evaluating and reversing technological dependence before it becomes irreversible. The analysis draws on documented sovereignty assessments conducted across national governments, financial institutions, healthcare systems, and educational organisations in multiple countries.
Part One: The Architecture of Dependency
Technology dependency is not a single condition. It is a graduated architecture — a structure constructed one decision at a time, through mechanisms that appear individually benign but accumulate into structural constraint. Understanding this architecture is the prerequisite for sovereignty-based governance.
The first mechanism is standardisation capture. When a provider establishes its definitions, protocols, and risk models as the default, it shapes the governance environment in ways that make divergence progressively more costly. A health ministry that adopts a diagnostic platform’s classification system does not merely adopt a tool — it adopts a worldview. The platform’s definitions of relevant factors, risk thresholds, and acceptable outcomes become the ministry’s definitions. Over time, the capacity to evaluate alternatives diminishes, because evaluation requires a reference frame, and the reference frame has been supplied by the platform.
The second mechanism is capability displacement. When an AI system performs analytical, decision-support, or monitoring functions that were previously performed by human professionals, those professionals lose practice. Their skills atrophy. The institution simultaneously gains a dependency on the system and loses the capacity to operate without it. This is the dependence trap: the system becomes necessary not because it was always necessary, but because the alternatives were allowed to degrade through disuse.
The third mechanism is contractual entrenchment. Providers embed dependency in contracts through proprietary formats, export restrictions, mandatory data access clauses, and pricing structures that make exit economically punishing. These provisions are rarely contentious at the point of contract signing, because the full cost of dependency is not yet visible. The cost becomes visible only when the institution attempts to exercise the option to leave — and discovers that the option was structurally foreclosed by the very agreement that provided access to the technology.
The fourth mechanism is narrative capture. Providers invest heavily in the narrative that their platforms represent progress, that resistance constitutes backwardness, and that sovereignty concerns are obstacles to efficiency. This narrative operates through every channel available: consultant advisory relationships, industry press coverage, conference sponsorship, and executive education programmes. The effect is to make the adoption decision appear not merely reasonable but virtuous — a demonstration of modernity rather than a governance choice that requires evaluation.
Part Two: How Dependency Manifests
Dependency does not announce itself. It accumulates silently, through individual decisions that each seem proportionate to their immediate context. A finance ministry adopts a tax analytics platform for efficiency. A customs authority implements an AI screening system for throughput. A university deploys a plagiarism detection tool for academic integrity. Each decision is reasonable in isolation. Each contributes to a profile that, when viewed in aggregate, reveals an architecture of dependence that no single decision was intended to construct.
The following five-domain Sovereignty Test Matrix provides a structured mechanism for assessing current dependency position. Score each domain from one (complete dependence) to five (complete sovereignty).
| Domain | Score (1-5) | What It Assesses | Key Questions |
|---|---|---|---|
| 1. Data Sovereignty | ___ | Location, access, and control of data | Where does data reside? Who controls it? Who can access it? |
| 2. System Architecture | ___ | Openness and portability of technical infrastructure | Are systems proprietary or open? Can they be migrated? |
| 3. Human Oversight | ___ | Retention of critical human analytical capacity | Do internal teams retain the skills to evaluate and govern the system independently? |
| 4. Contractual Governance | ___ | Exit provisions, data clauses, and pricing controls | Can the institution leave the provider? At what cost? |
| 5. Alternative Development | ___ | Existence of viable alternative systems or builders | Are alternative providers or domestic development options available? |
Scoring interpretation: A total score below ten indicates critical dependency requiring immediate governance intervention. A score between eleven and fifteen signals significant sovereignty risk that demands a structured remediation plan. A score between sixteen and twenty suggests moderate sovereignty with specific vulnerability domains requiring targeted improvement. A score above twenty demonstrates a robust sovereignty position, though continuous evaluation remains essential as the technological landscape evolves.
Red Flag Checklist
If three or more of the following apply to any single AI deployment, that deployment requires immediate sovereignty assessment:
☐ The institution cannot describe exactly what data the system processes or where that data travels.
☐ No one in the organisation could reconstruct the analytical model if the provider withdrew.
☐ The contract prevents using alternative systems concurrently without penalty.
☐ The system is described as critical infrastructure in internal documents but has no documented exit plan.
☐ The institution would require more than twelve months to replace the system.
☐ Key decisions made by the system are not reviewable by human staff with the requisite expertise.
☐ The contract includes mandatory data access clauses under foreign jurisdiction.
☐ Switching costs exceed twenty percent of the original deployment budget.
☐ No domestic or alternative provider has been evaluated as a potential replacement.
☐ The institution cannot function without the system for more than forty-eight hours.
The cost of divergence — international non-compliance, correspondent banking withdrawal, credit rating impact — was engineered to be prohibitive. This is the mechanism by which governance becomes capture: not through force, but through cost structures that make sovereign choice economically irrational.
Part Three: The TEE Method Framework for Dependency Resolution
The TEE Method — Test, Evaluate, Evolve — provides a structured process for assessing dependency and developing sovereignty strategies. Each phase addresses a different dimension of the dependency problem, and each produces documented outputs that inform subsequent phases.
TEST
Testing interrogates the dependency. It begins with a comprehensive inventory of all AI systems in use or under consideration. For each system, the following questions must be answered:
• What data does this system access, and where does that data reside?
• Who controls access to the system’s outputs and underlying models?
• What happens to the institution if the provider terminates the relationship?
• What human expertise currently exists within the institution to perform the system’s functions independently?
• What contractual provisions govern data export, system modification, and competitive development?
The test phase produces a Dependency Profile — a documented map of each system’s position across the five domains of the Sovereignty Test Matrix. This profile is the foundation for all subsequent governance action.
EVALUATE
Evaluation interrogates the strategic alignment between the system’s capabilities and the institution’s sovereignty objectives. The key evaluative questions are:
• Does this system serve the institution’s strategic interests, or does it serve primarily the provider’s interests?
• Does the system create value that the institution retains, or value that flows predominantly outward to the provider?
• Does the institution have sufficient leverage to negotiate improved sovereignty terms if they are not currently included?
• Would investing in domestic or alternative capability generate greater long-term sovereignty value than continuing the current dependency?
• Is there a viable exit strategy, and what would it cost to execute that strategy under current conditions?
Evaluation produces a Strategic Assessment — a structured analysis of each system’s contribution to or erosion of sovereignty. Systems that produce net sovereignty benefit are candidates for expanded deployment with enhanced governance frameworks. Systems that produce net erosion require remediation strategies that reduce dependency and restore institutional capacity.
EVOLVE
Evolution is the implementation phase. It encompasses five interlocking strategies that address the four mechanisms of dependency formation:
• Diversification: Deploy multiple providers across AI systems to prevent any single provider from becoming irreplaceable.
• Domestic capability: Invest in internal analytical teams that can evaluate, modify, and in some cases replace AI systems.
• Open-format mandates: Require that all new AI procurement supports open data formats and API standards that enable migration.
• Exit testing: Conduct annual exit readiness assessments for all critical systems. Each system must have a documented, tested exit protocol.
• Governance embedding: Ensure that sovereignty assessment is a mandatory component of all technology procurement decisions, not an optional review.
Sovereignty is not a condition. It is a practice — a continuous process of testing, evaluating, and evolving that requires institutional discipline and political will. The sovereign institution does not arrive at sovereignty. It works toward it every time it makes a technology decision.
Part Four: Actionable Steps
The following action plan provides week-by-week guidance for institutions seeking to assess and address technology dependency. The plan is designed for implementation by governance teams that include representatives from technology, legal, operational, and strategic functions.
Week One: Inventory
Conduct a comprehensive inventory of all AI systems in use or under active consideration. Document each system’s purpose, data flows, provider, contract status, and existing exit provisions. This inventory must be complete before any assessment can be meaningful.
Week Two: Scoring
Apply the Sovereignty Test Matrix to each system. Score each domain from one to five. Calculate aggregate scores and identify systems scoring below twelve. These are priority systems for immediate governance intervention.
Week Three: Red Flag Review
Conduct the Red Flag Checklist for all priority systems. Identify systems where three or more red flags apply. These systems require emergency remediation planning — they represent acute dependency risk that cannot be deferred.
Week Four: Strategic Assessment
Evaluate each system’s alignment with institutional strategic objectives. Document which systems generate net sovereignty benefit and which generate net erosion. Develop preliminary remediation plans for systems in the erosion category.
Week Five: Capability Building
Initiate internal capability building. Identify the analytical skills required to govern current and planned AI systems. Begin training programmes, recruitment processes, or partnership agreements that build domestic governance capacity.
Week Six: Contract Review
Conduct a legal review of all active AI contracts. Identify provisions that create dependency lock-in, restrict competitive development, or contain mandatory foreign-law data access clauses. Document required modifications and initiate renegotiation.
Week Seven: Exit Planning
Develop documented exit plans for all systems scoring below fifteen on the Sovereignty Test Matrix. Each exit plan must identify alternatives, estimated timelines, resource requirements, and migration procedures.
Week Eight: Governance Integration
Integrate sovereignty assessment into the institution’s standard technology procurement framework. Ensure that no new AI deployment can proceed without completing the Sovereignty Test Matrix and securing approved exit planning. Establish quarterly review cycles.
Every dependency begins with a decision that seemed reasonable at the time. Sovereignty begins with the decision to examine the architecture of those decisions — not to condemn the decisions, but to ensure that subsequent decisions are made with full awareness of their cumulative consequences.
The Question Revisited
What does technology dependency cost?
It costs the capacity to govern. It costs the skills that communities and institutions once held. It costs the leverage that independent capability provides in negotiations with powerful providers. It costs the resilience that comes from knowing that alternatives exist. And when dependency becomes structural — when the cost of exit exceeds the budget of the institution that depends — the cost is not permanently irreversible, but so large that it will typically not be paid. The institution learns to manage its dependency rather than resolve it. It negotiates from weakness. It accepts the provider’s definitions of value. It surrenders the capacity to say no.
But dependency is not destiny. It is a condition that can be assessed, mapped, and systematically reversed through the disciplined application of the TEE Method. The first step is testing — not as a compliance exercise, but as a sovereign act. The second is evaluation — not as an audit, but as a strategic reorientation. The third is evolution — not as an adjustment, but as a transformation of governance culture that places sovereign capability at the centre of institutional purpose.
The sovereign institution asks: Where are we dependent? What does it cost? And what will we build to change that position?
Those questions — and the discipline to answer them honestly — constitute the beginning of every genuine sovereignty strategy.
This article draws on the TEE Method framework from SOVEREIGN: Who Owns the Future? by Tonisha Tagoe (Chapter 13: Standardisation as Risk, Chapter 15: From Users to Builders).
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.
The TEE Method provides a structured process for evaluating dependency across all five domains. Each domain assessment produces a score from one to five, generating a composite sovereignty profile that guides remediation prioritisation. Domains scoring below three require immediate intervention. Domains scoring three to four require medium-term capability building. Domains scoring five represent strengths that can be leveraged to support capability development in weaker domains. The composite scoring mechanism ensures that remediation resources are directed to the areas of greatest need while recognising that sovereignty improvement is an ecosystem-wide process, not a series of isolated fixes.