THE SCENARIO
A multilateral forum convenes to negotiate a global AI governance framework. The drafting committee is composed of representatives from five AI-producing nations. The remaining one hundred eighty-eight member states are invited to “consult” on the final text. The framework establishes risk categories, compliance standards, and certification regimes that reflect the regulatory architectures of the drafting nations. It creates a tiered compliance structure: nations that adopt the framework as written receive “trusted partner” status; nations that adapt it to local conditions face “enhanced scrutiny.”
Within three years, the framework has become the prerequisite for participation in global AI supply chains. Nations that have not adopted it find their AI procurement options restricted, their research partnerships blocked, their access to compute corridors limited. The framework was never ratified by their legislatures. It was never subjected to democratic debate. It became binding through the architecture of dependency.
The nations that drafted it call this “harmonisation.” The nations that live under it call it something else.
The Question
How does a sovereign nation engage in global AI governance without surrendering its governance autonomy to frameworks drafted by and for AI-producing powers?
Part One: The Architecture of Multilateral Capture
Multilateral forums are not neutral spaces. They are arenas where power is exercised, alliances are formed, and outcomes are shaped by the relative influence of participants. The TEE Method acknowledges this reality and provides guidance for nations seeking to advance sovereignty interests in these arenas. The prevailing narrative presents multilateral AI governance as a technical exercise in harmonisation — agreeing on definitions, aligning standards, coordinating risk classifications. This narrative serves the interests of those who benefit from the status quo.
The drafting problem is structural. The nations that draft multilateral AI governance frameworks are overwhelmingly AI-producing nations. Their perspectives, assumptions, and interests are embedded in the language, structure, and priorities of the frameworks they produce. A framework drafted by a nation whose primary interest is promoting AI innovation will look different from a framework drafted by a nation whose primary interest is governing AI dependency. Both perspectives are legitimate. But the TEE Method insists that both must be represented in the drafting process — and that nations recognise the power inherent in drafting.
Consider what happens when a global AI governance framework is drafted exclusively by nations that export AI. The framework will prioritise: cross-border data flows (to enable model training), light-touch regulation (to accelerate deployment), mutual recognition of certifications (to reduce compliance costs for exporters), and innovation-friendly sandbox regimes (to speed time-to-market). These are not neutral choices. They are choices that advantage the drafters. The framework will not prioritise: data localisation (which constrains exporters), mandatory sovereignty assessments (which slow deployments), technology transfer requirements (which erode competitive advantage), or exit provisions (which reduce lock-in). These omissions are not accidental. They are the shape of exporter interest.
The dependent nations — the majority — are invited to “consult.” Consultation is not negotiation. Consultation means: here is the text, tell us your concerns, we will consider them. Negotiation means: here are our red lines, here are your red lines, let us find the agreement that respects both. The TEE Method demands that dependent nations recognise the difference — and refuse to confuse consultation with sovereignty.
The architecture of multilateral capture operates through three mechanisms. First, agenda control: the issues that matter to exporters are on the agenda; the issues that matter to the dependent are “technical details” for working groups. Second, language control: the definitions, categories, and standards are written in the exporters’ regulatory vocabulary; the dependent must translate their concerns into a language designed to exclude them. Third, implementation control: the frameworks are implemented through the exporters’ certification bodies, the exporters’ audit firms, the exporters’ technical assistance programmes. The dependent implements someone else’s rules — often paying for the privilege.
A framework drafted by exporters serves exporters. A framework drafted by the dependent serves the dependent. The TEE Method demands that the dependent be in the room when the draft is written.
Part Two: The Sovereignty Caucus — A Structural Intervention
The TEE Method proposes a practical mechanism for multilateral engagement: the Sovereignty Caucus — an informal coalition of nations that share a commitment to sovereign AI governance and coordinate their positions in multilateral forums. This is not a negotiating bloc in the traditional sense. It is a governance infrastructure. The Caucus performs three functions that transform multilateral participation from reactive to strategic.
Function One: Pre-Negotiation Position Development
Before major multilateral negotiations, the Caucus develops a shared position paper that identifies common interests, key demands, and red lines. This is not a wish list. It is a sovereignty assessment. Each member nation conducts a TEE National Readiness Index assessment. The aggregate reveals the coalition’s collective sovereignty position: where they are strong, where they are vulnerable, where they must hold the line, where they can trade. The position paper is not static. It evolves as the negotiation evolves. But it exists before the first plenary session — which means the Caucus enters the room with a sovereign position, not a blank page.
The position paper addresses the full spectrum of multilateral AI governance: data governance (localisation, portability, cross-border flow conditions), regulatory recognition (mutual recognition conditions, equivalence criteria, sovereign carve-outs), certification regimes (who certifies, what standards, appeal mechanisms), technology transfer (mandatory vs voluntary, scope, enforcement), compute access (allocation principles, sovereign reserves, pricing transparency), and dispute resolution (jurisdiction, standing, enforcement). Each issue is mapped to the National Readiness Index dimensions. The Caucus does not negotiate from aspiration. It negotiates from assessed capacity.
Function Two: Coordinated Intervention
During negotiations, Caucus members coordinate interventions. When a drafting nation proposes language that erodes data sovereignty, multiple Caucus members raise the objection — not as isolated complaints, but as a structured alternative. When a certification regime is proposed that privileges exporters’ compliance architectures, the Caucus presents a counter-proposal: a tiered recognition system that respects sovereign regulatory diversity. When a “harmonisation” clause is introduced that would require regulatory convergence, the Caucus proposes a “compatibilisation” alternative: interoperability without uniformity. The power of coordination is not rhetorical. It is procedural. Multilateral forums operate on consensus and compromise. A coordinated coalition that speaks with one voice on sovereignty provisions becomes a necessary partner in any consensus.
Coordination requires discipline. The Caucus agrees on: a lead negotiator for each agenda item, a shared language bank for sovereignty provisions, a rapid response protocol for surprise proposals, a whipping mechanism for critical votes, and a media strategy that frames sovereignty as legitimate governance — not obstruction. The Caucus does not seek to block all agreements. It seeks to ensure that every agreement respects the sovereignty floor: the minimum governance autonomy a nation must retain to be a sovereign participant rather than a managed dependency.
Function Three: Parallel Implementation
The most powerful function of the Caucus is what happens after the forum adjourns. Multilateral agreements are implemented nationally. The Caucus coordinates implementation — not to subvert the agreement, but to ensure that national implementation preserves the sovereignty protections the Caucus negotiated. This means: shared legislative templates for data sovereignty Acts, shared procurement regulations that embed TEE testing requirements, shared regulatory guidance that interprets framework provisions in sovereignty-preserving ways, and shared technical assistance for building the domestic capacity that makes sovereignty operational.
Parallel implementation is where sovereignty becomes real. A multilateral agreement that permits data localisation “where necessary” is implemented by one nation as a broad sovereign right and by another as a narrow exception. The Caucus ensures that its members implement the broad reading. A framework that requires “mutual recognition of certifications” is implemented by one nation as automatic acceptance and by another as a sovereign assessment process. The Caucus ensures its members implement the assessment process. The agreement is the same. The implementation is sovereignty.
The Caucus also builds the infrastructure for sustained coordination. A shared legal drafting unit. A shared technical assessment team. A shared intelligence function that monitors vendor lobbying, regulatory capture, and framework drift. A shared capacity-building programme that rotates personnel across member nations — building both capability and solidarity. The Sovereignty Caucus is not a meeting. It is an institution.
Multilateral engagement without a sovereign position is not participation. It is acquiescence.
Part Three: Diagnosing the Multilateral Position — The National Readiness Index
A nation cannot improve its multilateral position until it understands its current position. The TEE Method provides the National Readiness Index — a structured assessment across seven dimensions, each weighted to reflect its sovereignty impact. This is not a ranking exercise. It is a diagnostic instrument. The Index is conducted by a sovereign body — not by foreign consultants, not by vendor-funded assessors, not by the multilateral secretariat. The entity that assesses you shapes the assessment. The TEE Method insists: assess yourself first.
| Dimension | Weight | What It Measures | Sovereignty Indicator |
|---|---|---|---|
| Governance Architecture | 20% | Existence and authority of national AI governance structures | Can the nation make and enforce sovereign AI decisions? |
| Regulatory Framework | 15% | Completeness and enforceability of AI-specific legislation | Does the law reflect sovereign priorities or imported templates? |
| Technical Capacity | 15% | Domestic AI research, development, and deployment capability | Can the nation build, not just buy? |
| Workforce Sovereignty | 15% | Retention, development, and deployment of domestic AI talent | Does the nation own its expertise? |
| Data Sovereignty | 15% | Control over national data assets and governance of cross-border flows | Does the nation control its data destiny? |
| Research Capacity | 10% | Sovereign research agenda-setting and funding autonomy | Does the nation define its own research questions? |
| Infrastructure Independence | 10% | Control over compute, connectivity, and critical digital infrastructure | Can the nation operate without foreign infrastructure? |
Each dimension is scored 1-5. The weighted aggregate produces a National Readiness Score out of 100. A score below 40 indicates a nation that will be a rule-taker in multilateral forums. A score above 70 indicates a nation that can shape outcomes. Most nations score between 30 and 55. The gap is not ambition — it is institutional investment.
The Index must be recalculated annually. The TEE Method requires that the assessment be published — not the detailed scoring, but the aggregate position and the trajectory. Transparency creates accountability. A nation that publishes its National Readiness Score and its year-on-year change creates a sovereign benchmark that cannot be ignored in multilateral forums. The nation that hides its assessment surrenders the narrative to those who will assess it from the outside.
Each dimension deserves deeper examination. Governance Architecture is not merely the existence of an AI ministry or agency. It is the authority of that body to veto procurements, to mandate testing, to enforce data sovereignty, to coordinate across ministries, and to report directly to the head of government or parliament. Regulatory Framework is not the passage of an AI Act. It is the completeness of the regulatory chain: primary legislation, secondary regulations, regulatory guidance, enforcement mechanisms, and judicial review pathways — all reflecting sovereign priorities. Technical Capacity is not the number of AI engineers. It is the capacity to train foundation models, to audit third-party models, to operate sovereign compute, and to innovate at the architectural level. Workforce Sovereignty is not the production of AI graduates. It is the retention of those graduates, the alignment of their work with national priorities, and the protection of their expertise from extraction. Data Sovereignty is not data localisation alone. It is the governance of cross-border flows, the right to audit foreign processors, and the capacity to enforce data rights extraterritorially. Research Capacity is not national research spending. It is the autonomy to set research agendas, the ownership of research outputs, and the translation of research into sovereign capability. Infrastructure Independence is not ownership of data centres. It is the resilience of compute supply, the sovereignty of connectivity, and the hardness of critical digital infrastructure against coercion.
Red Flag Checklist: Is Your Nation a Rule-Taker?
If three or more apply, the nation is not preparing for multilateral engagement — it is preparing to be governed by it:
- The nation has no dedicated AI governance authority with legislative mandate.
- AI policy is developed by a committee that includes vendor representatives but no sovereignty advocates.
- The nation’s multilateral positions are drafted by foreign consultants or adopted from donor templates.
- No domestic AI procurement regulation exists — or it exempts “trusted partner” nations.
- Data localisation requirements have been waived for foreign AI providers under trade pressure.
- The national AI strategy was written with significant input from foreign AI companies.
- Compute infrastructure is 80% foreign-owned or foreign-located.
- The nation has never submitted a sovereign position paper to a multilateral AI forum.
- AI talent retention rate is below 50% at the five-year mark.
- No domestic AI research agenda exists independent of foreign funding priorities.
- The nation’s AI regulatory framework is a translated version of a foreign framework.
This checklist is not exhaustive. It is a starting point for the sovereign conversation that every government should be having — but rarely does. The absence of red flags does not prove sovereignty. The presence of red flags proves its absence.
Part Four: The Hundred-Day Multilateral Sovereignty Plan
The TEE Method provides a hundred-day plan for heads of state and ministers seeking to transform their nation’s multilateral position. This plan does not require massive budgets or technical expertise. It requires political will, governance discipline, and the questions that the TEE Method provides. A minister who executes this plan will know more about their ministry’s AI sovereignty position in one hundred days than most ministers learn in an entire tenure.
| Phase | Timeframe | Objective | Key Actions | Deliverable |
|---|---|---|---|---|
| Phase 1: Sovereignty Audit | Days 1-14 | Establish baseline; identify capture points | Conduct National Readiness Index; map all foreign AI dependencies; audit multilateral commitments | Sovereignty Baseline Report |
| Phase 2: Institutional Foundation | Days 15-40 | Create governance architecture | Establish AI Sovereignty Authority; mandate TEE testing for all procurement; legislate data sovereignty | Legal & institutional framework |
| Phase 3: Caucus Formation | Days 41-70 | Build multilateral coalition | Identify sovereign-aligned partners; negotiate Caucus charter; develop shared position template | Sovereignty Caucus operational |
| Phase 4: Position Deployment | Days 71-100 | Enter multilateral arena with sovereign position | Submit position papers; coordinate interventions; launch parallel implementation track | First sovereign multilateral intervention |
Phase 1 is the most neglected. Nations rush to “engage” without knowing what they are engaging for. The Sovereignty Baseline Report is not a public document. It is an internal decision tool. It maps every foreign AI system in government, every data flow crossing borders, every multilateral commitment that constrains sovereignty, every procurement pipeline that deepens dependency. It answers the question: if this system were withdrawn tomorrow, what would fail? If this data were accessed by a foreign power tomorrow, what would be exposed? If this multilateral commitment were enforced tomorrow, what sovereignty would be lost?
The Baseline Report must be produced by a team that combines technical expertise with governance authority. It cannot be outsourced to consultants — the act of producing it is itself a sovereignty exercise. The team needs: access to all government AI procurement records, authority to request data flow maps from every ministry, mandate to assess multilateral commitments against sovereign interests, and direct reporting line to the head of government. The report is classified. Its findings drive the hundred-day plan.
Phase 2 creates the architecture that makes sovereignty operational. The AI Sovereignty Authority is not an advisory body. It is a regulatory gate. No AI system enters government without its certification. The TEE testing mandate is not a guideline. It is a procurement condition. The data sovereignty legislation is not a principle. It is a law with penalties. These three instruments — authority, mandate, law — are the tripod on which sovereign governance stands. Without them, sovereignty is a sentiment. With them, sovereignty is a system.
The Authority’s design matters. It must be independent of the procurement function — it cannot evaluate what it procured. It must have veto power — not advisory opinion. It must be resourced — not starved into irrelevance. It must be accountable — to parliament, not to the executive alone. The TEE testing mandate must specify: what systems are tested, who conducts the test, what the passing threshold is, what happens when a system fails, and how often re-testing occurs. The data sovereignty legislation must address: data localisation for critical sectors, cross-border flow conditions, foreign access disclosure requirements, and enforcement against extraterritorial demands.
Phase 3 recognises that no nation achieves multilateral sovereignty alone. The Sovereignty Caucus is not a talking shop. It is a coordination mechanism with a charter, a workplan, and deliverables. The charter defines: membership criteria (National Readiness Score above threshold, commitment to sovereign governance, willingness to coordinate), decision-making (consensus on red lines, majority on tactics), secretariat (rotating, hosted by member with highest technical capacity), and accountability (annual peer review of implementation). The workplan aligns with the multilateral calendar: position papers due before each forum, coordinated interventions scheduled, parallel implementation milestones tracked.
The Caucus begins with a small core — three to five nations that have completed Phases 1 and 2. It does not wait for universal membership. It grows by demonstration: the core demonstrates that coordinated sovereignty works; others join to access the coordination. The Caucus is not open to nations that refuse the Baseline Report, the Authority, or the Mandate. Sovereignty is the membership fee.
Phase 4 is the test. The nation enters the multilateral arena not as a supplicant but as a sovereign with a position. The first intervention is not a speech. It is a submitted text: a position paper, a proposed amendment, a draft article. The intervention is coordinated — multiple Caucus members submit aligned texts. The intervention is sustained — not one forum, but every forum. The intervention is implemented — parallel implementation begins immediately. The hundred-day plan does not end at day one hundred. It begins there.
The Question Revisited
How does a sovereign nation engage in global AI governance without surrendering its governance autonomy to frameworks drafted by and for AI-producing powers? It engages by preparing first. It engages by building the institutional capacity to have a sovereign position. It engages by coordinating with others who share the sovereignty commitment. It engages by bringing its own draft to the table — not by annotating someone else’s.
The TEE Method provides the framework. The Sovereignty Caucus provides the vehicle. The National Readiness Index provides the baseline. The hundred-day plan provides the timeline. But none of these operate without the decision: that sovereignty is not the price of participation — it is the precondition.
This article draws on the TEE Method framework from SOVEREIGN: Who Owns the Future?