This is an automated rejection. No LLM generated, heavily assisted/co-written, or otherwise reliant work.
Read full explanation
ALGORITHMIC SOVEREIGNTY AND GLOBAL
GOVERNANCE OF ARTIFICIAL INTELLIGENCE
A framework for the sovereignty of states and individuals in the algorithmic age
Academic Essay
2026
Area: Governance, Geopolitics and Political Philosophy of Artificial Intelligence
Algorithmic Sovereignty and Global AI Governance | 1 Abstract
This essay analyzes one of the most urgent and unresolved problems in contemporary global politics: the inability of states and individuals to exercise real sovereignty over their territories and lives when the artificial intelligence systems that structure their daily existence were designed, trained, and governed under foreign legal and cultural frameworks.
Building on that central problem, the essay constructs an original governance architecture composed of three levels: sovereign algorithmic infrastructures organized by geopolitical blocs, a neutral-by-design interoperability bridge, and an international mutual deterrence treaty that ensures said bridge cannot be weaponized as a tool of coercion. The essay argues that this proposal, though imperfect and transitional in nature, represents the most viable and politically realistic framework for protecting the algorithmic sovereignty of states and individuals within the current geopolitical context.
1. Introduction: The Problem of Sovereignty in the Algorithmic Age
Artificial intelligence does not respect borders. A language model trained in California can be used simultaneously by millions of people in Lagos, Buenos Aires, Berlin, or Jakarta. Yet regulation does have borders: it is national, or at most regional. This structural contradiction between the global nature of AI systems and the territorial nature of legal frameworks constitutes the core of the problem this essay sets out to analyze.
The question is not entirely new in philosophical terms. Classical sovereignty, as formulated in the Peace of Westphalia in 1648, rests on the assumption that a State can exercise effective control over what happens within its territory. That assumption was
Algorithmic Sovereignty and Global AI Governance | 2 extended, with increasing difficulty, to airspace, the maritime domain, and more recently to cyberspace. Today we face a fourth dimension of sovereignty that none of those existing theoretical or institutional frameworks is equipped to manage: algorithmic sovereignty.
Algorithmic sovereignty can be defined as the real capacity of a State or individual to control the automated decision-making systems that structure their functioning as a society. When those systems are designed, trained, and governed from another country, that sovereignty is compromised in ways that are not always visible, but that are profoundly real.
2. The Central Problem: A Chain of Overlapping Sovereignties
The most common framing of this problem tends to oversimplify it: it is described as a situation where an AI based in another country applies that country's laws to users in the recipient country. But that description underestimates the real complexity of the phenomenon.
When a user interacts with an AI system, they are not subject to a single foreign jurisdiction. They are subject to multiple overlapping sovereignties simultaneously, and in most cases without their knowledge or consent:
The model was trained under the laws and culture of the country where the developing company is headquartered. The servers on which it runs may be located in a third country with different privacy and security regulations. The data used for training comes from users across dozens of countries, each with their own regulatory frameworks. And the output that reaches the end user has been mediated by all of those layers before appearing on their screen.
This overlapping of sovereignties operates at two qualitatively distinct levels. The first is visible and relatively easy to identify: it occurs when an AI explicitly applies legal restrictions or censorship from its country of origin to users in other countries. The second
Algorithmic Sovereignty and Global AI Governance | 3 is more subtle and philosophically more significant: the AI does not need to impose any explicit law on the user to affect their sovereignty. It does so through the implicit values embedded in its training, the cultural biases it contains, the categories through which it organizes the world, and the options it makes visible or invisible. This process of normative modeling does not appear in any contract or regulation, but it silently shapes how users think, what options they perceive as possible, and what decisions they ultimately make.
The question is not whether an AI has values, but whether those values are visible, contestable, and democratically legitimate for those who live under their influence.
3. Why Existing Governance Frameworks Are Insufficient
3.1 National Regulation and the Brussels Effect
The most advanced governance framework currently in existence is the European Union's Artificial Intelligence Act. Its risk-based classification architecture represents a significant normative achievement. However, its territorial scope is its fundamental limitation: it applies within the EU and, through the so-called Brussels Effect, it compels global companies wishing to operate in that market to comply with its standards. But it has no binding power outside its borders, and its enforcement is inevitably mediated by the political and economic interests of the powers that make up the bloc.
3.2 The Internet Governance Model Is Not Transferable A frequent response to the problem of global AI governance consists in invoking the model that worked, with its limitations, for the internet: technical bodies like ICANN, multilateral forums like the IGF, and open protocols like TCP/IP that enable interoperability without imposing specific cultural values.
That model is not transferable to AI for a fundamental structural reason: the internet was designed with a decentralized architecture from its very origin. Frontier AI, by contrast, is concentrated in a very small number of companies that require computational resources — processing power, data, and specialized human capital — accessible only to the
Algorithmic Sovereignty and Global AI Governance | 4 wealthiest countries and corporations in the world. There is no algorithmic equivalent of the email server that any organization can operate independently. Power is concentrated in a way that fundamentally changes the nature of the governance problem.
3.3 The International Institutional Vacuum
No international body with binding power over artificial intelligence currently exists. Existing initiatives — the GPAI, UNESCO's recommendation on AI ethics, the security summits inaugurated at Bletchley Park in 2023 — are statements of intent without enforcement mechanisms. The structure of the UN Security Council, with its veto system for the five permanent members, makes any agreement affecting the interests of the major technological powers structurally impossible.
SWIFT, the international interbank payments system, offers a brutal historical precedent: presented for decades as neutral infrastructure, its use as a political weapon against Russia in 2022 demonstrated that all global infrastructure is, ultimately, an instrument of power for whoever controls it.
4. An Architecture of Distributed Algorithmic Sovereignty
Faced with the inadequacy of existing frameworks, this essay proposes a governance architecture composed of three articulated levels, each with a specific function and its own resilience mechanisms.
4.1 First Level: Sovereign Algorithmic Infrastructures by Bloc The first level of the proposed architecture is also the most urgent in terms of national security: each country or geopolitical bloc must develop and control its own algorithmic infrastructure for the management of critical data. This includes, at minimum, financial systems, defense logistics, health infrastructure, and identity and property registries.
The reason for this urgency is not merely normative but strategic. An AI embedded in a country's financial infrastructure whose owning company is headquartered in a State with which that country enters into diplomatic or military conflict represents a first-order
Algorithmic Sovereignty and Global AI Governance | 5 national security vulnerability. The State that owns the AI does not need to declare conventional war: it can silently degrade the system's performance, withdraw technical support, or introduce perturbations that are difficult to attribute and easy to deny. This dimension of the problem — which we might call algorithmic infrastructure colonialism —
transforms the question of algorithmic sovereignty from a philosophical debate into a matter of state survival.
Smaller countries that lack the resources to develop their own algorithmic infrastructure will need, during this transition, to lease capabilities from allied countries or blocs of which they form part. This partially reproduces existing asymmetries, but in a substantially less dangerous way than the current dependency, where dozens of countries rely on systems managed from potentially antagonistic jurisdictions. Full sovereignty has never been equally accessible to all States; what this architecture seeks is to limit the damage and reduce systemic exposure.
4.2 Second Level: A Neutral-by-Design Interoperability Bridge The second level addresses the need for communication between the different sovereign infrastructures. Absolute algorithmic autarky is neither desirable nor feasible: States and blocs need systems that enable cross-border financial transfers, exchange of critical information, and coordination during global emergencies.
For this purpose, the essay proposes the design of a distributed-architecture interoperability bridge, whose technical governance is organized in such a way that no single bloc has unilateral capacity to control it. The fundamental principle is structural neutrality by design: not because no actor has interests, but because the interests of different actors limit each other through consensus mechanisms and shared auditing.
This bridge would operate under the following technical and political principles: open source code auditable by all participating blocs; distributed governance where no bloc holds a controlling majority; the capacity of each bloc to disconnect unilaterally without collapsing its own infrastructure; and a modification protocol requiring multi-bloc consensus for any substantial change.
Algorithmic Sovereignty and Global AI Governance | 6 4.3 Third Level: The Algorithmic Mutual Deterrence Treaty The third level is what gives the entire architecture real teeth: an international treaty establishing the mutual deterrence mechanism for the interoperability bridge.
The treaty's logic is simple and deliberately symmetrical: any State or bloc that uses the bridge as a coercion instrument against another — whether by disconnecting it, degrading its service, or manipulating information flows — will be automatically disconnected from the system at the same time. Not as the consequence of a vote in an international committee susceptible to political capture, but as an immediate and unappealable structural consequence inscribed in the technical architecture of the system itself.
This is the central innovation of the proposal relative to existing international governance frameworks. All of them share the same Achilles' heel: they depend on the most powerful actors voluntarily complying with norms they themselves helped draft. The algorithmic mutual deterrence mechanism inverts that logic: compliance is not voluntary but structural. The incentives of all actors — including the most powerful — are aligned with respect for the system, because the alternative is their own exclusion.
The analogy with the Mutually Assured Destruction doctrine of the Cold War is explicit and intentional. No nuclear power launched a first strike for decades not because it had stopped having conflicting interests, but because the consequence of doing so was its own destruction. The proposed treaty applies that same geometry of incentives to the algorithmic domain: unilateral disconnection is technically possible, but its cost is identical to what it seeks to inflict on the other party.
5. Irreducible Tensions and Limits of the Proposal
A rigorous academic essay cannot present its proposal without honestly confronting its weaknesses. The proposed architecture has at least four tensions that cannot be fully resolved within its own logic.
The first is the problem of the constitutive neutrality of the bridge. Designing a distributed governance infrastructure requires making technical decisions that always embed value
Algorithmic Sovereignty and Global AI Governance | 7 choices. Who participates in its initial design, under what protocols, with what definition of consensus: all of those choices incorporate particular cultural and geopolitical perspectives. Structural neutrality by design mitigates this problem but does not eliminate it.
The second is the risk of false positives. The automatic disconnection mechanism presupposes that it is possible to identify with certainty who initiated a coercive action. But a malicious actor could deliberately provoke the appearance of an attack in order to trigger mutual disconnection between two rival blocs. This would require a technical verification protocol prior to automatic activation, which reintroduces the human factor that the architecture sought to minimize.
The third is the problem of contagion effects. The disconnection of two conflicting blocs could affect third countries whose financial communications depended on the bridge to interact with both blocs. The treaty would need to contemplate emergency corridors and bypass routes for countries not involved in the conflict.
The fourth is the deepest and has no technical solution: the structural inequality in the capacity to build sovereign algorithmic infrastructure. The countries most in need of algorithmic protection are, paradoxically, those least capable of building it. This architecture does not resolve that asymmetry; it constrains it, but perpetuates it in a different form.
6. The Shared Sovereign Moral Code: A Normative Complement to the Technical Architecture
The three-level architecture described in the previous section resolves the problem of infrastructure dependency, but does not address the normative dimension of the problem: the fact that AI systems incorporate cultural and ideological values that affect users regardless of who controls the infrastructure.
For this level of the problem, a complementary mechanism is proposed: the development of a shared sovereign moral code, generated from the algorithmic synthesis of the
Algorithmic Sovereignty and Global AI Governance | 8 common denominators across the constitutions of participating countries, and validated by an international committee of broad representation.
The logic of this proposal inverts the problem of value imposition. Rather than having a group of actors with particular interests decide which values should govern global AI systems, the AI itself is used as a synthesis instrument to identify the principles that already exist as tacit consensus among sovereign nations, as codified in their
constitutional texts.
This mechanism has limitations the essay cannot ignore. Constitutions are historically and culturally situated documents, interpreted in natural language that an AI system will inevitably read through the biases of its own training. States are not cultural monoliths: a constitutional synthesis could reflect the perspective of drafting elites more than that of the majority. And in authoritarian contexts, adaptation to the local constitutional framework could become a tool of political control.
For those reasons, the shared moral code is proposed not as a binding framework but as a voluntary opt-in reference: AI systems that adopt it declare so explicitly; those that do not are not required to do so either. Users and States have the right to know under what normative framework the system they use operates, and to choose not to use it if that framework is unacceptable to them. This logic respects the real political pluralism of the world without imposing a version of universalism that is itself not neutral.
7. Conclusions: Sovereignty in the Twenty-First Century
This essay has argued that sovereignty, as conceived at Westphalia and as extended to cyberspace, is insufficient to protect the autonomy of states and individuals in the age of artificial intelligence. Existing governance frameworks — whether national, regional, or the timid international attempts that have emerged — share the same structural flaw: they depend on the willingness of the most powerful actors to comply, and those actors are precisely the ones with the most to gain from the absence of effective regulation.
Algorithmic Sovereignty and Global AI Governance | 9 In response to that insufficiency, the essay has proposed a three-level governance architecture: sovereign algorithmic infrastructures by bloc, a neutral-by-design interoperability bridge, and a mutual deterrence treaty that makes compliance structural rather than voluntary. This architecture is not perfect. It perpetuates some asymmetries, introduces new tensions, and does not resolve the problem of inequality in the capacity to build sovereign infrastructure. But it is politically more realistic than any proposal that rests on the goodwill of actors or on international bodies whose legitimacy is compromised by the very power relations they claim to regulate.
Twenty-first century sovereignty is not measured solely in territory, military power, or gross domestic product. It is also measured in algorithmic capacity: in the real possibility that a State manages the automated decision-making systems that sustain its functioning, and in the real possibility that an individual understands, contests, and eventually rejects the normative frameworks under which the systems that structure their life operate.
Countries that fail to develop that capacity, or to build alliances that grant them access to it, will be structural dependents of those who do: in times of peace, through the silent imposition of foreign values; in times of conflict, through the vulnerability of their critical infrastructure. The architecture proposed in this essay is a temporary patch, but it is the most robust patch that current political logic allows us to build. And beginning to build it is more urgent than most public debates about artificial intelligence seem to recognize.
References and Recommended Reading
Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press.
Dafoe, A. (2018). AI Governance: A Research Agenda. Future of Humanity Institute, University of Oxford.
Gebru, T. (2019). Oxford Handbook on AI Ethics: Race and Gender. Oxford University Press.
Krasner, S. D. (1999). Sovereignty: Organized Hypocrisy. Princeton University Press.
Algorithmic Sovereignty and Global AI Governance | 10 Risse, M. (2019). Human Rights and Artificial Intelligence: An Urgently Needed Agenda. Human Rights Quarterly, 41(1), 1–16.
Scharre, P. (2023). Four Battlegrounds: Power in the Age of Artificial Intelligence. W.W. Norton & Company.
European Union (2024). Regulation (EU) 2024/1689 of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act).
Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs.
Algorithmic Sovereignty and Global AI Governance | 11
ALGORITHMIC SOVEREIGNTY AND GLOBAL
GOVERNANCE OF ARTIFICIAL INTELLIGENCE
A framework for the sovereignty of states and individuals in the algorithmic age
Academic Essay
2026
Area: Governance, Geopolitics and Political Philosophy of Artificial Intelligence
Algorithmic Sovereignty and Global AI Governance | 1
Abstract
This essay analyzes one of the most urgent and unresolved problems in contemporary global politics: the inability of states and individuals to exercise real sovereignty over their territories and lives when the artificial intelligence systems that structure their daily existence were designed, trained, and governed under foreign legal and cultural frameworks.
Building on that central problem, the essay constructs an original governance architecture composed of three levels: sovereign algorithmic infrastructures organized by geopolitical blocs, a neutral-by-design interoperability bridge, and an international mutual deterrence treaty that ensures said bridge cannot be weaponized as a tool of coercion. The essay argues that this proposal, though imperfect and transitional in nature, represents the most viable and politically realistic framework for protecting the algorithmic sovereignty of states and individuals within the current geopolitical context.
Keywords: algorithmic sovereignty, AI governance, infrastructure warfare, mutual deterrence, neutral interoperability, algorithmic colonialism.
1. Introduction: The Problem of Sovereignty in the Algorithmic Age
Artificial intelligence does not respect borders. A language model trained in California can be used simultaneously by millions of people in Lagos, Buenos Aires, Berlin, or Jakarta. Yet regulation does have borders: it is national, or at most regional. This structural contradiction between the global nature of AI systems and the territorial nature of legal frameworks constitutes the core of the problem this essay sets out to analyze.
The question is not entirely new in philosophical terms. Classical sovereignty, as formulated in the Peace of Westphalia in 1648, rests on the assumption that a State can exercise effective control over what happens within its territory. That assumption was
Algorithmic Sovereignty and Global AI Governance | 2
extended, with increasing difficulty, to airspace, the maritime domain, and more recently to cyberspace. Today we face a fourth dimension of sovereignty that none of those existing theoretical or institutional frameworks is equipped to manage: algorithmic sovereignty.
Algorithmic sovereignty can be defined as the real capacity of a State or individual to control the automated decision-making systems that structure their functioning as a society. When those systems are designed, trained, and governed from another country, that sovereignty is compromised in ways that are not always visible, but that are profoundly real.
2. The Central Problem: A Chain of Overlapping Sovereignties
The most common framing of this problem tends to oversimplify it: it is described as a situation where an AI based in another country applies that country's laws to users in the recipient country. But that description underestimates the real complexity of the phenomenon.
When a user interacts with an AI system, they are not subject to a single foreign jurisdiction. They are subject to multiple overlapping sovereignties simultaneously, and in most cases without their knowledge or consent:
The model was trained under the laws and culture of the country where the developing company is headquartered. The servers on which it runs may be located in a third country with different privacy and security regulations. The data used for training comes from users across dozens of countries, each with their own regulatory frameworks. And the output that reaches the end user has been mediated by all of those layers before appearing on their screen.
This overlapping of sovereignties operates at two qualitatively distinct levels. The first is visible and relatively easy to identify: it occurs when an AI explicitly applies legal restrictions or censorship from its country of origin to users in other countries. The second
Algorithmic Sovereignty and Global AI Governance | 3
is more subtle and philosophically more significant: the AI does not need to impose any explicit law on the user to affect their sovereignty. It does so through the implicit values embedded in its training, the cultural biases it contains, the categories through which it organizes the world, and the options it makes visible or invisible. This process of normative modeling does not appear in any contract or regulation, but it silently shapes how users think, what options they perceive as possible, and what decisions they ultimately make.
The question is not whether an AI has values, but whether those values are visible, contestable, and democratically legitimate for those who live under their influence.
3. Why Existing Governance Frameworks Are Insufficient
3.1 National Regulation and the Brussels Effect
The most advanced governance framework currently in existence is the European Union's Artificial Intelligence Act. Its risk-based classification architecture represents a significant normative achievement. However, its territorial scope is its fundamental limitation: it applies within the EU and, through the so-called Brussels Effect, it compels global companies wishing to operate in that market to comply with its standards. But it has no binding power outside its borders, and its enforcement is inevitably mediated by the political and economic interests of the powers that make up the bloc.
3.2 The Internet Governance Model Is Not Transferable A frequent response to the problem of global AI governance consists in invoking the model that worked, with its limitations, for the internet: technical bodies like ICANN, multilateral forums like the IGF, and open protocols like TCP/IP that enable interoperability without imposing specific cultural values.
That model is not transferable to AI for a fundamental structural reason: the internet was designed with a decentralized architecture from its very origin. Frontier AI, by contrast, is concentrated in a very small number of companies that require computational resources — processing power, data, and specialized human capital — accessible only to the
Algorithmic Sovereignty and Global AI Governance | 4
wealthiest countries and corporations in the world. There is no algorithmic equivalent of the email server that any organization can operate independently. Power is concentrated in a way that fundamentally changes the nature of the governance problem.
3.3 The International Institutional Vacuum
No international body with binding power over artificial intelligence currently exists. Existing initiatives — the GPAI, UNESCO's recommendation on AI ethics, the security summits inaugurated at Bletchley Park in 2023 — are statements of intent without enforcement mechanisms. The structure of the UN Security Council, with its veto system for the five permanent members, makes any agreement affecting the interests of the major technological powers structurally impossible.
SWIFT, the international interbank payments system, offers a brutal historical precedent: presented for decades as neutral infrastructure, its use as a political weapon against Russia in 2022 demonstrated that all global infrastructure is, ultimately, an instrument of power for whoever controls it.
4. An Architecture of Distributed Algorithmic Sovereignty
Faced with the inadequacy of existing frameworks, this essay proposes a governance architecture composed of three articulated levels, each with a specific function and its own resilience mechanisms.
4.1 First Level: Sovereign Algorithmic Infrastructures by Bloc The first level of the proposed architecture is also the most urgent in terms of national security: each country or geopolitical bloc must develop and control its own algorithmic infrastructure for the management of critical data. This includes, at minimum, financial systems, defense logistics, health infrastructure, and identity and property registries.
The reason for this urgency is not merely normative but strategic. An AI embedded in a country's financial infrastructure whose owning company is headquartered in a State with which that country enters into diplomatic or military conflict represents a first-order
Algorithmic Sovereignty and Global AI Governance | 5
national security vulnerability. The State that owns the AI does not need to declare conventional war: it can silently degrade the system's performance, withdraw technical support, or introduce perturbations that are difficult to attribute and easy to deny. This dimension of the problem — which we might call algorithmic infrastructure colonialism —
transforms the question of algorithmic sovereignty from a philosophical debate into a matter of state survival.
Smaller countries that lack the resources to develop their own algorithmic infrastructure will need, during this transition, to lease capabilities from allied countries or blocs of which they form part. This partially reproduces existing asymmetries, but in a substantially less dangerous way than the current dependency, where dozens of countries rely on systems managed from potentially antagonistic jurisdictions. Full sovereignty has never been equally accessible to all States; what this architecture seeks is to limit the damage and reduce systemic exposure.
4.2 Second Level: A Neutral-by-Design Interoperability Bridge The second level addresses the need for communication between the different sovereign infrastructures. Absolute algorithmic autarky is neither desirable nor feasible: States and blocs need systems that enable cross-border financial transfers, exchange of critical information, and coordination during global emergencies.
For this purpose, the essay proposes the design of a distributed-architecture interoperability bridge, whose technical governance is organized in such a way that no single bloc has unilateral capacity to control it. The fundamental principle is structural neutrality by design: not because no actor has interests, but because the interests of different actors limit each other through consensus mechanisms and shared auditing.
This bridge would operate under the following technical and political principles: open source code auditable by all participating blocs; distributed governance where no bloc holds a controlling majority; the capacity of each bloc to disconnect unilaterally without collapsing its own infrastructure; and a modification protocol requiring multi-bloc consensus for any substantial change.
Algorithmic Sovereignty and Global AI Governance | 6
4.3 Third Level: The Algorithmic Mutual Deterrence Treaty The third level is what gives the entire architecture real teeth: an international treaty establishing the mutual deterrence mechanism for the interoperability bridge.
The treaty's logic is simple and deliberately symmetrical: any State or bloc that uses the bridge as a coercion instrument against another — whether by disconnecting it, degrading its service, or manipulating information flows — will be automatically disconnected from the system at the same time. Not as the consequence of a vote in an international committee susceptible to political capture, but as an immediate and unappealable structural consequence inscribed in the technical architecture of the system itself.
This is the central innovation of the proposal relative to existing international governance frameworks. All of them share the same Achilles' heel: they depend on the most powerful actors voluntarily complying with norms they themselves helped draft. The algorithmic mutual deterrence mechanism inverts that logic: compliance is not voluntary but structural. The incentives of all actors — including the most powerful — are aligned with respect for the system, because the alternative is their own exclusion.
The analogy with the Mutually Assured Destruction doctrine of the Cold War is explicit and intentional. No nuclear power launched a first strike for decades not because it had stopped having conflicting interests, but because the consequence of doing so was its own destruction. The proposed treaty applies that same geometry of incentives to the algorithmic domain: unilateral disconnection is technically possible, but its cost is identical to what it seeks to inflict on the other party.
5. Irreducible Tensions and Limits of the Proposal
A rigorous academic essay cannot present its proposal without honestly confronting its weaknesses. The proposed architecture has at least four tensions that cannot be fully resolved within its own logic.
The first is the problem of the constitutive neutrality of the bridge. Designing a distributed governance infrastructure requires making technical decisions that always embed value
Algorithmic Sovereignty and Global AI Governance | 7
choices. Who participates in its initial design, under what protocols, with what definition of consensus: all of those choices incorporate particular cultural and geopolitical perspectives. Structural neutrality by design mitigates this problem but does not eliminate it.
The second is the risk of false positives. The automatic disconnection mechanism presupposes that it is possible to identify with certainty who initiated a coercive action. But a malicious actor could deliberately provoke the appearance of an attack in order to trigger mutual disconnection between two rival blocs. This would require a technical verification protocol prior to automatic activation, which reintroduces the human factor that the architecture sought to minimize.
The third is the problem of contagion effects. The disconnection of two conflicting blocs could affect third countries whose financial communications depended on the bridge to interact with both blocs. The treaty would need to contemplate emergency corridors and bypass routes for countries not involved in the conflict.
The fourth is the deepest and has no technical solution: the structural inequality in the capacity to build sovereign algorithmic infrastructure. The countries most in need of algorithmic protection are, paradoxically, those least capable of building it. This architecture does not resolve that asymmetry; it constrains it, but perpetuates it in a different form.
6. The Shared Sovereign Moral Code: A Normative Complement to the Technical Architecture
The three-level architecture described in the previous section resolves the problem of infrastructure dependency, but does not address the normative dimension of the problem: the fact that AI systems incorporate cultural and ideological values that affect users regardless of who controls the infrastructure.
For this level of the problem, a complementary mechanism is proposed: the development of a shared sovereign moral code, generated from the algorithmic synthesis of the
Algorithmic Sovereignty and Global AI Governance | 8
common denominators across the constitutions of participating countries, and validated by an international committee of broad representation.
The logic of this proposal inverts the problem of value imposition. Rather than having a group of actors with particular interests decide which values should govern global AI systems, the AI itself is used as a synthesis instrument to identify the principles that already exist as tacit consensus among sovereign nations, as codified in their
constitutional texts.
This mechanism has limitations the essay cannot ignore. Constitutions are historically and culturally situated documents, interpreted in natural language that an AI system will inevitably read through the biases of its own training. States are not cultural monoliths: a constitutional synthesis could reflect the perspective of drafting elites more than that of the majority. And in authoritarian contexts, adaptation to the local constitutional framework could become a tool of political control.
For those reasons, the shared moral code is proposed not as a binding framework but as a voluntary opt-in reference: AI systems that adopt it declare so explicitly; those that do not are not required to do so either. Users and States have the right to know under what normative framework the system they use operates, and to choose not to use it if that framework is unacceptable to them. This logic respects the real political pluralism of the world without imposing a version of universalism that is itself not neutral.
7. Conclusions: Sovereignty in the Twenty-First Century
This essay has argued that sovereignty, as conceived at Westphalia and as extended to cyberspace, is insufficient to protect the autonomy of states and individuals in the age of artificial intelligence. Existing governance frameworks — whether national, regional, or the timid international attempts that have emerged — share the same structural flaw: they depend on the willingness of the most powerful actors to comply, and those actors are precisely the ones with the most to gain from the absence of effective regulation.
Algorithmic Sovereignty and Global AI Governance | 9
In response to that insufficiency, the essay has proposed a three-level governance architecture: sovereign algorithmic infrastructures by bloc, a neutral-by-design interoperability bridge, and a mutual deterrence treaty that makes compliance structural rather than voluntary. This architecture is not perfect. It perpetuates some asymmetries, introduces new tensions, and does not resolve the problem of inequality in the capacity to build sovereign infrastructure. But it is politically more realistic than any proposal that rests on the goodwill of actors or on international bodies whose legitimacy is compromised by the very power relations they claim to regulate.
Twenty-first century sovereignty is not measured solely in territory, military power, or gross domestic product. It is also measured in algorithmic capacity: in the real possibility that a State manages the automated decision-making systems that sustain its functioning, and in the real possibility that an individual understands, contests, and eventually rejects the normative frameworks under which the systems that structure their life operate.
Countries that fail to develop that capacity, or to build alliances that grant them access to it, will be structural dependents of those who do: in times of peace, through the silent imposition of foreign values; in times of conflict, through the vulnerability of their critical infrastructure. The architecture proposed in this essay is a temporary patch, but it is the most robust patch that current political logic allows us to build. And beginning to build it is more urgent than most public debates about artificial intelligence seem to recognize.
References and Recommended Reading
Crawford, K. (2021). Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press.
Dafoe, A. (2018). AI Governance: A Research Agenda. Future of Humanity Institute, University of Oxford.
Gebru, T. (2019). Oxford Handbook on AI Ethics: Race and Gender. Oxford University Press.
Krasner, S. D. (1999). Sovereignty: Organized Hypocrisy. Princeton University Press.
Algorithmic Sovereignty and Global AI Governance | 10
Risse, M. (2019). Human Rights and Artificial Intelligence: An Urgently Needed Agenda. Human Rights Quarterly, 41(1), 1–16.
Scharre, P. (2023). Four Battlegrounds: Power in the Age of Artificial Intelligence. W.W. Norton & Company.
European Union (2024). Regulation (EU) 2024/1689 of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act).
Zuboff, S. (2019). The Age of Surveillance Capitalism. PublicAffairs.
Algorithmic Sovereignty and Global AI Governance | 11