1. Introduction
Artificial intelligence (AI) has become an embedded feature of contemporary governance, influencing everything from social welfare delivery to transnational trade regulation. Its capacity to analyse vast quantities of data and automate decision-making processes has placed it squarely at the intersection of law, technology, and sovereignty. Yet, as governments assert increasing control over digital infrastructures, they also expose deep fault lines in international law, particularly concerning territoriality, jurisdiction, and accountability (Leiter, Reference Leiter2020, p. 3). Traditional doctrines were designed for a world of physical borders and discrete state actions. Algorithmic systems, by contrast, operate through networked and cross-border logics, often diffusing state responsibility and eroding clarity over which jurisdiction governs their operation.
The notion of AI sovereignty has therefore emerged as a critical concept within the evolving architecture of international law. It refers to a state’s authority to regulate and shape the development, deployment, and impact of AI systems within its jurisdiction (Srinath, Reference Srinath2024, p. 72). In practice, however, AI sovereignty often extends beyond domestic law. States now seek to influence global technology standards, data flows, and algorithmic ethics in ways that resemble classical struggles over territorial sovereignty (Hung, Reference Hung2025, p. 119). These developments pose a dual challenge: first, how can international law accommodate such expansive assertions of digital jurisdiction, and second, how can accountability be ensured when algorithmic decisions transcend territorial boundaries?
This article interrogates AI sovereignty as both a legal construct and a site of normative tension. It argues that the governance of AI cannot be confined within the traditional categories of public international law; rather, it requires the evolution of hybrid frameworks that combine sovereign control with cross-border cooperation. The study is grounded in comparative analysis and jurisprudential inquiry, focusing on how Asian legal systems, specifically China, India, and Association of Southeast Asian Nations (ASEAN) jurisdictions, have crafted distinct models of digital sovereignty. These models are contrasted with the European Union’s (EU) supranational rights-based framework and the United States’ market-oriented approach, revealing fundamental differences in how legal orders perceive the balance between innovation, rights, and state authority (Fleming, Reference Fleming2025).
The article proposed three major interventions. First, the article contends that the idea of territoriality must be reconceptualised to account for the spatial and operational realities of digital infrastructures. Second, it advocates the creation of cross-border accountability mechanisms capable of addressing algorithmic harms that occur across jurisdictions. Third, it calls for the establishment of treaty-based frameworks that institutionalise cooperation on AI without undermining sovereign agency (Crootof, Reference Crootof2018, p. 577). Together, these measures are aimed at reshaping the normative foundations of international law to better reflect the algorithmic conditions of global governance.
Methodologically, the article adopts a comparative socio-legal approach, drawing upon theories of legal pluralism and global constitutionalism to explore how overlapping legal orders interact in the digital realm. The Asian region is presented not merely as a passive subject of global regulation but as an active producer of normative models that challenge Western assumptions about law and technology. By situating Asia at the centre of this discourse, the article highlights the region’s potential to redefine international legal debates on sovereignty and accountability in the digital era.
Finally, the study seeks to contribute to a broader understanding of how algorithmic governance reshapes the international legal order. It suggests that the future of global law will depend on reconciling sovereign control with transnational cooperation, and that Asia’s evolving legal and political experiments provide essential insights into this reconciliation.
2. AI sovereignty and digital territoriality
The concept of AI sovereignty represents the contemporary evolution of a much older legal idea: the right of a state to control activities within its territory and regulate conduct that affects its population. In the digital era, this notion has expanded to include control over data, algorithms, and digital infrastructure elements that often transcend physical borders and operate in virtual domains. AI sovereignty therefore denotes a state’s claim to jurisdictional authority over AI systems, encompassing both their domestic deployment and their transnational impact (Srinath, Reference Srinath2024, p. 72). The assertion of such authority has profound implications for international law, as it redefines the limits of sovereignty and complicates the spatial logic of territorial jurisdiction.
Initially, digital governance debates revolved around data sovereignty, which concerned the state’s capacity to regulate the collection, processing, and storage of data generated within its borders. Over time, however, scholars and policymakers have recognised that data control alone is insufficient. Algorithms, particularly those employing machine learning, transform raw data into automated decisions with real-world consequences for individuals, institutions, and markets (Avbelj, Reference Avbelj2024, p. 1083). The state’s interest has thus shifted from mere data protection to algorithmic regulation, encompassing how AI systems are designed, trained, and deployed.
As Vasanthika Srinath (Reference Srinath2024, p. 73) explains, algorithmic sovereignty involves not only data localisation or cybersecurity measures but also the capacity to shape algorithmic logic in accordance with domestic legal and ethical norms. For example, India’s evolving digital policy framework now requires that algorithms used in governance be auditable and explainable, reflecting an attempt to align technological systems with constitutional values such as equality and accountability. In this sense, algorithmic sovereignty functions as an instrument of constitutional self-definition in the digital sphere, allowing states to articulate what forms of AI are acceptable within their normative order. Similarly, in China, the government has adopted an expansive interpretation of cyber sovereignty, claiming authority over all aspects of the country’s digital ecosystem, including algorithms used by private platforms (Hung, Reference Hung2025, p. 121). The Cybersecurity Law (2017), Data Security Law (2021), and Personal Information Protection Law (2021) collectively impose extensive obligations on entities operating AI systems within Chinese cyberspace. These statutes enable the state to regulate not only the flow of data but also the ethical and operational design of algorithms. Such regulation exemplifies a broader political and legal vision in which AI governance is inseparable from national security and ideological control (Leiter, Reference Leiter2020, p. 4).
The shift from data to algorithmic sovereignty reflects a structural transformation in global governance. States now perceive AI as both a tool of economic competitiveness and a potential vector of geopolitical vulnerability. This duality has intensified efforts to nationalise algorithmic infrastructures through policies on data localisation, indigenous AI research, and restrictions on foreign software, mirroring the industrial policies of the twentieth century but framed in the language of digital independence (Fleming, Reference Fleming2025).
The emergence of AI sovereignty has forced scholars and policymakers to reconsider one of the foundational doctrines of international law, i.e. territoriality. Traditionally, jurisdiction was tied to the physical location of conduct or effects: what occurred within a state’s borders fell under its legal control. However, in the algorithmic age, the connection between location and legal authority is increasingly tenuous. Data are stored across multiple jurisdictions, algorithms process information on distributed servers, and decisions generated in one country may affect individuals halfway across the world.
Andrea Leiter (Leiter, Reference Leiter2020, p. 6) notes that cyberspace and data flows have “decoupled the exercise of sovereignty from physical geography,” producing a new spatial logic that traditional territorial concepts cannot easily accommodate. Digital infrastructures are deterritorialised. It exists simultaneously everywhere and nowhere. This condition creates what some scholars call “virtual territoriality,” a domain in which states attempt to map jurisdictional claims onto flows of information and computation rather than physical territory.
For instance, the European Union’s General Data Protection Regulation (GDPR) has extended the notion of territorial jurisdiction beyond the EU’s borders by applying its provisions to any entity processing data about EU residents, regardless of where the entity is located. This extraterritorial reach demonstrates how digital territoriality can be reconceptualised as a function of effects and personal nexus, rather than mere physical presence. Conversely, China’s model emphasises internal territorial sovereignty, asserting that all digital activities conducted within its cyber infrastructure, whether by domestic or foreign actors, fall under national jurisdiction (Hung, Reference Hung2025, p. 125).
India’s approach lies somewhere in between. The Digital Personal Data Protection Act (2023) incorporates data localisation requirements for sensitive personal information, signifying a territorial claim over data flows originating from Indian citizens, while also engaging in international negotiations to facilitate data transfer frameworks. This dual strategy reflects what Unnikrishnan (Reference Unnikrishnan2025) calls “pragmatic sovereignty,” a balancing act between protectionism and global participation.
Thus, territoriality in the digital age no longer signifies exclusive physical control but represents a multi-layered and networked concept of jurisdiction. States define their territorial reach not only by geography but also by technological capability, control over data nodes, and influence over digital norms. The challenge for international law is to develop principles that can mediate these overlapping territorial claims without undermining the open architecture of cyberspace.
The assertion of AI sovereignty has important implications for the structure of the international legal order. Traditionally, sovereignty served as both the foundation and the limit of international law: states consented to be bound by rules but retained the ultimate authority within their borders. The rise of algorithmic governance disrupts this equilibrium. As digital technologies blur the lines between domestic and international domains, states are compelled to cooperate on issues that were once purely internal, such as data protection, cybersecurity, and AI ethics (Avbelj, Reference Avbelj2024, p. 1088).
This reconfiguration has prompted calls for new normative frameworks that reconcile national sovereignty with global accountability. Fleming (Reference Fleming2025) observes that divergent models of digital governance, such as the EU’s rights-based approach, China’s security-oriented model, and the United States’ market-driven framework, reflect competing visions of sovereignty itself. If these remain uncoordinated, they risk producing fragmented regulatory ecosystems, often referred to as “digital blocs,” which could impede international cooperation and exacerbate inequality between technologically advanced and developing nations.
From a jurisprudential standpoint, AI sovereignty reveals the tension between legal pluralism and constitutional universalism. Legal pluralists view the multiplicity of digital regimes as inevitable and potentially beneficial, allowing experimentation and contextual adaptation (Leiter, Reference Leiter2020, p. 8). Constitutional universalists, on the other hand, argue that the transnational nature of AI necessitates a shared legal foundation, grounded in human rights and global norms (Avbelj, Reference Avbelj2024, p. 1087). The challenge for international law lies in striking a balance between these positions, acknowledging the legitimacy of diverse national approaches while promoting minimum universal standards to prevent harm and ensure accountability.
3. Asian approaches to digital sovereignty
Asian engagements with algorithmic governance do not reflect a unified or homogeneous conception of digital sovereignty. Rather, they embody distinct constitutional traditions, governance philosophies, and developmental priorities. For analytical clarity, this section examines three representative Asian approaches: China’s state-centric model of cyber sovereignty, India’s constitutional and infrastructure-driven approach to algorithmic governance, and ASEAN’s cooperative and pluralist framework. Together, these models demonstrate how sovereignty is operationalised through different legal techniques while remaining central to the regulation of AI in the region.
3.1. China: state-centric cyber sovereignty
China has been the most explicit and ideologically coherent proponent of cyber sovereignty, which it defines as the right of states to regulate the Internet and digital technologies within their borders according to national laws and cultural values. President Xi Jinping’s statement that “no country should interfere in another’s internal affairs in cyberspace” epitomises this doctrine, extending the principle of non-intervention to digital domains (Hung, Reference Hung2025, p. 122). This vision positions the state as the primary architect and regulator of algorithmic systems, asserting political authority over both domestic technology companies and foreign platforms operating in Chinese cyberspace.
China’s legislative framework reinforces this stance. The Cybersecurity Law (2017), Data Security Law (2021), and Personal Information Protection Law (2021) collectively create a dense web of regulatory obligations that govern data processing, cross-border transfers, and algorithmic transparency (Leiter, Reference Leiter2020, p. 5). Under these laws, companies must store certain categories of data within Chinese territory and submit to government audits. In 2022, the Algorithmic Recommendation Management Provisions went further, requiring online platforms to disclose how their algorithms influence user behaviour and mandating content moderation that aligns with state-defined “social values” (Hung, Reference Hung2025, p. 124).
This approach serves several policy goals. Domestically, it enhances control over the information ecosystem, aligning digital regulation with China’s broader objectives of political stability and social harmony. Economically, it seeks to foster technological self-reliance, reducing dependency on Western technology by promoting indigenous innovation in AI, semiconductors, and data analytics. Geopolitically, it asserts China’s vision of a sovereign Internet, resisting the dominance of Western liberal norms and presenting an alternative governance model for the Global South (Fleming, Reference Fleming2025).
However, critics argue that China’s cyber sovereignty model risks deepening global digital fragmentation, creating parallel ecosystems that hinder interoperability and cross-border collaboration (Leiter, Reference Leiter2020, p. 8). Proponents, on the other hand, view it as a legitimate extension of Westphalian sovereignty into the digital domain, ensuring that national interests are not subordinated to transnational corporate power or external political influence. Whether seen as defensive or assertive, China’s stance undeniably reshapes the conversation on digital governance, compelling the international community to confront questions of sovereignty, accountability, and rights in the algorithmic age.
3.2. India: constitutional accountability and digital public infrastructure
India’s engagement with digital sovereignty is rooted in a different historical and constitutional context. Rather than asserting unilateral control over cyberspace, India has pursued a pragmatic and developmentalist approach, using digital governance as a means to enhance state capacity and public service delivery. Initiatives such as Digital India, Aadhaar, and the Unified Payments Interface (UPI) exemplify how algorithmic systems are being integrated into welfare distribution, financial inclusion, and identity management (Srinath, Reference Srinath2024, p. 74).
India’s trajectory towards algorithmic sovereignty began with the Supreme Court’s landmark decision in Justice K.S. Puttaswamy v Union of India (2017), which recognised privacy as a fundamental right under Article 21 of the Constitution. This judgement established the constitutional foundation for subsequent data and AI governance frameworks, mandating that any interference with informational autonomy must satisfy tests of legality, necessity, and proportionality (Unnikrishnan, Reference Unnikrishnan2025). The Court’s reasoning implied that sovereignty over data and algorithms must be exercised in ways consistent with constitutional principles, a perspective that differentiates India’s model from more security-driven approaches like China’s.
The Digital Personal Data Protection Act (2023) builds on this foundation by granting individuals rights to access, correction, and erasure of personal data while imposing obligations on entities that process such data (Unnikrishnan, Reference Unnikrishnan2025). The Act also empowers the government to notify categories of sensitive personal data that must be stored or processed within India, reflecting a moderated form of data localisation. Simultaneously, the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules (2021) require online intermediaries to ensure traceability of messages and remove unlawful content, signalling a growing assertion of state authority over digital platforms.
Yet, unlike China’s command-and-control model, India’s strategy seeks to reconcile sovereignty with open digital ecosystems. The development of India Stack, a set of interoperable public digital infrastructure components, demonstrates the government’s intent to create a foundational digital layer accessible to private and public actors alike (Unnikrishnan, Reference Unnikrishnan2025). This architecture supports both innovation and accountability, allowing the state to regulate without monopolising control. As Srinath (Reference Srinath2024, p. 77) notes, India’s emphasis on explainability and transparency in algorithmic decision-making reflects a constitutional vision of technological self-determination rather than techno-authoritarianism.
At the same time, India’s engagement in global digital diplomacy through forums such as the Global Partnership on Artificial Intelligence (GPAI) and the G20 Digital Economy Working Group illustrates its desire to influence international standards while maintaining national flexibility. The Indian model thus represents a hybrid sovereignty: assertive in protecting domestic interests, yet cooperative in shaping international norms.
3.3. ASEAN: cooperative and pluralist digital sovereignty
The ASEAN offers a third, more pluralistic approach to digital sovereignty. Unlike China or India, ASEAN as a regional bloc lacks a unified constitutional framework; its member states vary significantly in political systems, economic development, and technological capability. Consequently, ASEAN has pursued a cooperative, capacity-building approach to algorithmic governance, emphasising dialogue, inclusivity, and gradual standard setting rather than rigid regulation (Turobov, Reference Turobov2025).
ASEAN’s Digital Master Plan 2025 and the Guidelines on Artificial Intelligence Ethics and Governance (2023) reflect its pragmatic orientation. These documents prioritise regional integration and economic development, seeking to harmonise AI-related standards while respecting national sovereignty. As Turobov (Reference Turobov2025) observes, ASEAN’s strategy is characterised by “building bridges rather than walls,” positioning the region as a neutral interlocutor in the global contest between US- and China-led digital blocs. This diplomacy-first approach allows ASEAN to engage with multiple partners simultaneously; for instance, collaborating with Japan and the EU on data protection while participating in China’s Belt and Road Digital Initiative.
Domestically, ASEAN members exhibit considerable diversity. Singapore has developed a robust AI governance framework, including the Model AI Governance Framework (2020), which emphasises transparency, fairness, and accountability. Malaysia and Thailand have followed similar paths, integrating AI ethics into their national strategies. Other member states, such as Indonesia and Vietnam, are at earlier stages, focusing on capacity development and digital literacy. What unites these diverse trajectories is a commitment to balancing openness with sovereignty, avoiding binary alignments, and instead fostering regional consensus on responsible AI.
ASEAN’s pluralism is both its strength and its challenge. On one hand, it accommodates diverse political and developmental contexts, ensuring inclusivity. On the other hand, the absence of binding regional enforcement mechanisms limits its ability to implement uniform digital standards. Nonetheless, ASEAN’s “third way” offers a model for cooperative governance in a fragmented world, one that privileges dialogue, flexibility, and incremental harmonisation over rigid sovereignty claims (Turobov, Reference Turobov2025).
A comparison of these three models reveals both convergence and divergence in Asia’s understanding of digital sovereignty. China prioritises security and ideological control, India emphasises constitutional accountability and infrastructural capacity, and ASEAN promotes cooperation and inclusivity. Despite their differences, all three models share a recognition that digital governance must remain an attribute of sovereign authority. Yet, unlike the EU’s rights-based universalism or the US market individualism, Asian models situate sovereignty within contextual and developmental logics reflecting the historical realities of post-colonial state formation and economic modernisation (Avbelj, Reference Avbelj2024, p. 1089).
This comparative diversity underscores Asia’s centrality in global algorithmic governance. Far from being a passive recipient of Western regulatory norms, the region is generating alternative pathways for integrating sovereignty, technology, and law. These models invite the international community to reconceptualise governance not as a binary between national control and globalisation, but as a continuum of negotiated autonomy. As later sections argue, this pluralistic foundation is essential for designing cross-border accountability mechanisms and treaty frameworks that respect sovereign agency while ensuring global coherence.
4. Western approaches to algorithmic governance: the European Union and the United States
The contrasting trajectories of the EU and the United States represent two dominant paradigms of algorithmic governance in the Western world. The EU has adopted a rights-based and supranational model, viewing AI primarily through the lens of human dignity, privacy, and democratic accountability. The United States, conversely, has maintained a market-oriented and innovation-driven approach, privileging economic competitiveness and technological entrepreneurship over regulatory intervention. Together, these models shape global debates about digital sovereignty, extraterritorial regulation, and the ethical limits of AI (Fleming, Reference Fleming2025).
4.1. The European Union: supranational right-based regulation
The EU’s approach to AI governance is anchored in its broader constitutional project of protecting fundamental rights within a shared legal order. Since the adoption of the Charter of Fundamental Rights of the European Union (2000), the EU has positioned itself as a normative power that seeks to embed human rights standards into all aspects of digital regulation (Avbelj, Reference Avbelj2024, p. 1087). The cornerstone of this framework is the General Data Protection Regulation (GDPR) 2016/679, which has become a global benchmark for privacy and data governance.
The GDPR extends the EU’s regulatory reach beyond its territorial boundaries by applying its provisions to any entity that processes the personal data of EU residents, regardless of where the processing occurs. This extraterritorial application codified under Article 3(2) has effectively globalised the EU’s conception of privacy as a transnational right (Fleming, Reference Fleming2025). By compelling multinational corporations to comply with European data standards, the EU has transformed privacy protection into a form of digital sovereignty exercised through normative influence rather than coercive power.
Building upon this foundation, the EU has turned its attention to AI-specific regulation through the Artificial Intelligence Act (AI Act), formally adopted in 2024. The Act introduces a risk-based regulatory model, classifying AI systems into categories of unacceptable, high, limited, and minimal risk (European Commission, 2024). High-risk AI systems such as those used in biometric surveillance, employment screening, or credit scoring must comply with stringent requirements of transparency, accuracy, and human oversight. By contrast, low-risk systems are subject to lighter obligations. This calibrated approach reflects the EU’s commitment to fostering innovation while safeguarding rights and values.
The EU’s AI governance model is thus an extension of its legal philosophy of “constitutional pluralism,” the belief that supranational norms can coexist with national autonomy under a shared commitment to rights and rule of law (Avbelj, Reference Avbelj2024, p. 1088). The Digital Services Act (2022) and the Digital Markets Act (2022) further reinforce this ecosystem by regulating online platforms and promoting fair competition. Together, these instruments form a coherent framework for algorithmic accountability, where both private and public actors are bound by transparent, rights-based obligations.
However, this model is not without criticism. Scholars such as Hung (Reference Hung2025, p. 130) note that the EU’s reliance on legalistic regulation may slow innovation and disadvantage smaller states or companies unable to meet compliance costs. Moreover, the EU’s attempt to project its standards globally, sometimes referred to as the “Brussels Effect,” has raised questions about legitimacy and consent. Critics argue that by exporting its laws extraterritorially, the EU risks imposing a digital neocolonialism, particularly on developing countries that lack the capacity to comply (Fleming, Reference Fleming2025). Nonetheless, the EU’s framework remains a crucial experiment in reconciling technological change with constitutional order, and it continues to shape global discussions on algorithmic ethics and legal harmonisation.
4.2. The United States: market-oriented and decentralised governance
In contrast to the EU’s regulatory activism, the United States has traditionally adopted a market-oriented and decentralised approach to digital governance. Rooted in its liberal economic ideology, the US model privileges innovation, competition, and corporate self-regulation as primary drivers of technological advancement (Fleming, Reference Fleming2025). Until recently, federal regulation of AI and data protection has been fragmentary and sector-specific, reflecting a deep-seated belief that premature or excessive regulation could stifle innovation and economic growth.
At the federal level, there is no comprehensive privacy or AI legislation comparable to the GDPR or the EU AI Act. Instead, governance relies on a patchwork of statutes such as the Federal Trade Commission Act (1914), which prohibits unfair or deceptive business practices, and the Children’s Online Privacy Protection Act (1998). These laws provide a basic framework for consumer protection but fall short of addressing the complex ethical and legal issues posed by modern AI systems (Leiter, Reference Leiter2020, p. 6).
In response to growing public concern over algorithmic bias and data misuse, recent administrations have begun to explore soft-law instruments and executive initiatives. The Blueprint for an AI Bill of Rights (2022), issued by the White House Office of Science and Technology Policy, outlines five guiding principles: safe and effective systems, protection from algorithmic discrimination, data privacy, notice and explanation, and human alternatives. Although non-binding, the Blueprint represents an important normative shift towards rights-oriented AI governance. Similarly, the Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (2023) mandates federal agencies to ensure transparency, testing, and risk management in AI procurement and deployment (The White House, 2023).
At the subnational level, states such as California, Colorado, and Virginia have enacted their own privacy statutes, including the California Consumer Privacy Act (CCPA) and the California Privacy Rights Act (CPRA), which grant individuals rights over data access and correction. These laws signal a move towards decentralised regulatory pluralism, mirroring the federal structure of US governance (Fleming, Reference Fleming2025). Additionally, the National Institute of Standards and Technology (NIST) has published an AI Risk Management Framework (2023) that offers voluntary standards for transparency, fairness, and accountability.
Despite these efforts, the United States remains fundamentally market-centric in its digital philosophy. Its global influence derives not from law but from the dominance of its technology companies such as Google, Microsoft, and OpenAI, which export software, cloud infrastructure, and AI tools worldwide. Through contractual norms and technological standards, these corporations exercise a quasi-regulatory authority that rivals state power (Leiter, Reference Leiter2020, p. 9). This privatised governance structure contrasts sharply with the EU’s legalism and China’s statist control, highlighting a uniquely American vision of algorithmic order grounded in entrepreneurial freedom and innovation.
However, the laissez-faire approach also produces vulnerabilities. The absence of comprehensive regulation has led to inconsistent accountability standards and insufficient protection against algorithmic harms, including discrimination and misinformation. Moreover, the concentration of AI development within a few corporate entities has raised concerns about digital oligopolies, where private interests dictate global norms without democratic oversight (Avbelj, Reference Avbelj2024, p. 1091). These issues have prompted renewed calls for legislative reform, suggesting that the United States may soon move towards a hybrid model combining innovation incentives with human rights safeguards.
A defining feature of the US approach to algorithmic governance is the displacement of regulatory authority from the state to private technological actors. In practice, major technology firms exercise forms of quasi-sovereign power by setting algorithmic standards, moderating content, and determining access to digital markets across jurisdictions. This privatisation of governance complicates international law’s traditional state-centric framework, as regulatory influence is exercised extraterritorially through corporate infrastructure rather than formal jurisdiction. Consequently, algorithmic harms generated by US-based platforms often fall into accountability gaps, where neither domestic public law nor international legal mechanisms provide effective oversight. The US model thus exposes a central tension in global AI governance: when sovereignty is exercised through markets rather than law, international accountability becomes structurally difficult to enforce.
A comparison of the EU and US approaches reveals profound differences in legal philosophy but also important points of convergence. The EU’s model reflects a normative vision of global digital sovereignty, in which law functions as a vehicle for exporting human rights values. The US model, by contrast, embodies a functional vision of sovereignty rooted in market leadership and technological dominance. While Europe seeks to regulate through law, the United States governs through economic scale and innovation ecosystems.
Despite these divergences, both regions face common challenges. First, the rapid pace of technological innovation continually outstrips the capacity of law to adapt. Second, both struggle to establish cross-border accountability mechanisms for algorithmic harms that transcend national boundaries. Third, each model risks exporting its internal contradictions: the EU’s regulatory density may lead to over-compliance and innovation bottlenecks, while the United States’ corporate-led governance risks undermining public trust and global equity (Fleming, Reference Fleming2025).
In this context, Asian approaches, particularly the plural and adaptive models of China, India, and ASEAN, offer valuable lessons. Their emphasis on contextual sovereignty, infrastructural self-reliance, and cooperative engagement provides a counterpoint to Western binaries of state control versus market freedom. As subsequent sections argue, these Asian experiences can inform the development of hybrid international frameworks that reconcile sovereignty with accountability in the algorithmic age.
5. Theoretical framework: legal pluralism and global constitutionalism
The governance of AI through the prism of international law presents not merely a policy or technical challenge but a profound theoretical dilemma. The doctrines that traditionally structured the international legal order, such as sovereignty, jurisdiction, and accountability, were conceived for a world where law was territorially bounded and state authority was singular. In contrast, algorithmic governance unfolds across multiple legal, technological, and institutional layers, blurring the distinction between domestic and international, public and private. Two conceptual frameworks, legal pluralism and global constitutionalism, offer useful analytical tools for understanding this transformation and for imagining how law might evolve to address the complexities of AI sovereignty.
The theory of legal pluralism recognises that multiple legal systems, i.e. state law, international law, religious law, customary norms, and private governance regimes, often coexist and interact within the same social space. Applied to the digital context, legal pluralism challenges the assumption that sovereignty is singular or hierarchical. Instead, it posits that the digital domain is governed by a mosaic of normative orders, including the technical protocols of private companies, the ethical guidelines of international organisations, and the domestic laws of states (Leiter, Reference Leiter2020, p. 8).
In cyberspace, private actors exercise what has been called “functional sovereignty,” deriving authority not from state consent but from control over digital architectures. Corporations such as Google, Meta, and Microsoft establish rules for content moderation, data retention, and algorithmic transparency that have global reach and binding effect. These norms operate independently of, and often beyond, traditional state jurisdiction (Crootof, Reference Crootof2018, p. 579). This phenomenon underscores the argument advanced by legal pluralists like Boaventura de Sousa Santos that “the state is no longer the exclusive producer of legality” (de Sousa Santos, Reference de Sousa Santos1987, p. 288). In the realm of algorithmic governance, private code effectively becomes law, a reality captured by Lawrence Lessig’s famous dictum that “code is law” (Lessig, Reference Lessig1999, p. 6).
From a pluralist standpoint, the coexistence of multiple regulatory orders is not necessarily problematic; it may, in fact, enhance adaptability and contextual sensitivity. For instance, the decentralised model of AI governance in the United States allows for experimentation through soft-law instruments and state-level legislation. Similarly, ASEAN’s cooperative model enables incremental harmonisation without rigid supranational oversight (Turobov, Reference Turobov2025). These forms of normative plurality reflect the evolving nature of sovereignty in the digital age, no longer an exclusive prerogative of the state but a distributed phenomenon shared among various actors and institutions.
However, pluralism also generates conflict. Overlapping claims of authority often lead to jurisdictional collisions, as illustrated by disputes over data localisation or extraterritorial enforcement of the EU’s GDPR. In such cases, pluralism without coordination risks legal fragmentation and uncertainty (Avbelj, Reference Avbelj2024, p. 1085). Thus, while legal pluralism accurately describes the reality of algorithmic governance, it must be complemented by frameworks that promote coherence, accountability, and respect for universal values. This is where global constitutionalism provides a necessary counterbalance.
The concept of global constitutionalism seeks to extend constitutional principles such as the rule of law, separation of powers, and fundamental rights to the international legal order. Proponents argue that as global governance regimes gain authority and affect individual lives, they must also be subject to constitutional constraints akin to those that bind domestic governments (Avbelj, Reference Avbelj2024, p. 1088). In the digital era, where algorithms can influence elections, shape markets, and determine access to justice, the need for such constraints becomes especially urgent.
At its core, global constitutionalism envisions international law as an evolving constitutional framework for humanity, within which states, international organisations, and private actors operate under shared principles. For algorithmic governance, this means embedding transparency, accountability, and human rights into the design and operation of AI systems (Crootof, Reference Crootof2018, p. 580). The approach resonates with the Council of Europe’s Framework Convention on Artificial Intelligence (2024), which affirms that AI must respect human rights, democracy, and the rule of law across its lifecycle (Council of Europe, 2024). By articulating universal principles without prescribing specific technologies, the Convention exemplifies a constitutional response to technological disruption.
However, global constitutionalism remains contested. Critics argue that it risks imposing a Western-centric conception of rights and governance on diverse legal cultures, particularly those in Asia and the Global South (Fleming, Reference Fleming2025). Others question whether international institutions possess the democratic legitimacy required to function as constitutional authorities. As Krisch (Reference Krisch2010, p. 7) contends, the post-national order may require “constitutional pluralism rather than constitutional unity,” recognising the coexistence of multiple constitutional projects rather than a single global constitution.
In this light, global constitutionalism should not be understood as a universal blueprint but as a normative horizon, a set of shared commitments that guide the evolution of international law without erasing cultural and political diversity. For AI governance, these commitments might include the protection of human dignity, procedural fairness, and algorithmic transparency, principles that can be realised through diverse institutional arrangements.
The interaction between legal pluralism and global constitutionalism offers a productive framework for rethinking AI sovereignty. On one hand, pluralism acknowledges the multiplicity of actors and regimes shaping algorithmic governance; on the other hand, constitutionalism insists that these actors be bound by common principles of legitimacy and accountability. The challenge for international law is to design mechanisms that preserve diversity without anarchy, that is, a pluralism anchored in constitutional values.
In practical terms, this synthesis requires moving beyond rigid hierarchies of law. Instead of seeking a singular global regulator, international law can facilitate polycentric governance, in which states, corporations, and international bodies collaborate through overlapping yet coordinated mechanisms (Avbelj, Reference Avbelj2024, p. 1091). For example, treaty-based frameworks could set baseline principles of algorithmic fairness while allowing states to tailor domestic implementation according to local contexts. Similarly, soft-law instruments such as the OECD AI Principles or UNESCO’s Recommendation on the Ethics of Artificial Intelligence can complement binding obligations, providing flexibility while reinforcing shared norms.
This pluralist-constitutional approach aligns with Asia’s diverse experiences. China’s state-centric control, India’s constitutional accountability, and ASEAN’s cooperative pluralism each represent legitimate expressions of sovereignty within a global system. The task is not to homogenise these models but to ensure that they operate within an overarching normative framework that protects fundamental values and facilitates coordination. As the following sections propose, such a synthesis can be operationalised through three key interventions: reconceptualising digital territoriality, developing cross-border accountability mechanisms, and establishing treaty frameworks that embed cooperation without eroding sovereign agency.
6. Reconceptualising digital territoriality
The first necessary intervention in reimagining international law for the algorithmic age concerns the meaning of territoriality. For more than three centuries, territorial sovereignty has served as the organising principle of the international legal system: jurisdiction was presumed to follow geography. Yet in an era defined by data flows, distributed computing, and transnational algorithmic operations, the territorial metaphor has begun to disintegrate. The challenge for contemporary jurists is to articulate a form of digital territoriality that preserves legitimate state interests while recognising the networked and deterritorialised nature of AI systems.
Classical international law rests on the premise that sovereignty is exercised over a fixed geographic space and that legal authority derives from a state’s control over that space. This premise underpinned doctrines of jurisdiction—territorial, nationality-based, and protective—that continue to inform modern treaties and domestic statutes. However, as Andrea Leiter (Reference Leiter2020, p. 5) observes, “the spatial logic of data has unmoored legal authority from geography,” since digital activities occur simultaneously across multiple servers and jurisdictions. When an algorithm trained in Singapore analyses data collected in Kenya and delivers results to users in India, no single territorial nexus adequately captures the locus of regulation.
The problem is not merely technical; it is conceptual. Traditional jurisdictional tests such as where the act occurs or where the effect is felt lose coherence when data storage, processing, and decision-making are geographically disaggregated. In this context, the insistence on physical location can lead either to regulatory overreach, as when multiple states claim concurrent jurisdiction, or to regulatory voids, where none can enforce accountability. The rise of AI thus exposes the insufficiency of territoriality as the exclusive basis for legal authority (Avbelj, Reference Avbelj2024, p. 1085).
Despite these challenges, states have not relinquished the territorial paradigm; they have repurposed it for the digital sphere. Several models of digital jurisdiction have emerged.
6.1. The effects doctrine revisited
Borrowed from antitrust and securities law, the effects test attributes jurisdiction to a state when conduct abroad produces substantial effects within its territory. The EU’s GDPR exemplifies this logic, extending EU authority to any controller or processor handling EU residents’ data, irrespective of location. By framing data subjects rather than territory as the jurisdictional anchor, the EU reconceptualises space in functional terms (European Commission, 2016). Similar reasoning appears in consumer-protection jurisprudence in India, where courts have asserted jurisdiction over foreign e-commerce platforms serving Indian customers (Unnikrishnan, Reference Unnikrishnan2025). Although the effects doctrine provides a functional basis for extending jurisdiction in the digital context, its application to algorithmic systems raises unresolved concerns regarding proportionality and regulatory overlap. When multiple states simultaneously claim jurisdiction over the same AI system based on diffuse and indirect effects, the doctrine risks exacerbating legal uncertainty rather than resolving it. This underscores the need for coordinated jurisdictional principles capable of mediating overlapping sovereign claims.
6.2. Data localisation and digital borders
China and, increasingly, India have adopted data localisation mandates that require certain categories of data to be stored or processed domestically (Hung, Reference Hung2025, p. 124; Srinath, Reference Srinath2024, p. 73). These measures recreate physical borders within cyberspace, transforming data centres into instruments of sovereignty. Proponents argue that localisation enhances security and ensures compliance with national laws; critics contend that it fragments the Internet and restricts innovation. Nonetheless, localisation has become a powerful symbol of digital territorial control, an attempt to restore spatial certainty in an otherwise borderless medium.
6.3. Platform territoriality
A third, less formal model arises from the quasi-territorial power of large technology corporations. Through terms of service, content policies, and algorithmic governance, platforms create private jurisdictions that span the globe (Lessig, Reference Lessig1999, p. 6). Users often consent implicitly to these rules as a condition of participation. The result is a pluralistic order in which private digital territories coexist with public law regimes, sometimes in conflict and sometimes in symbiosis. Facebook’s Oversight Board and Google’s AI Ethics Guidelines illustrate how corporate entities now perform functions once reserved for states.
Together, these patterns indicate that territoriality has not disappeared but has become multilayered. Physical, functional, and normative territories overlap, producing what legal pluralists describe as a polycentric system of governance (Avbelj, Reference Avbelj2024, p. 1090).
A future-oriented theory of digital territoriality should aim to reconcile three objectives: (1) the protection of state sovereignty and legitimate public interests; (2) the prevention of jurisdictional conflict; and (3) the facilitation of transnational accountability.
6.3.1. Functional attachment tests
International law could develop functional attachment criteria to determine which state bears regulatory responsibility for an algorithmic system. Factors might include the location of data subjects, the primary site of system design, and the locus of decision-making authority. This multifactorial approach mirrors conflict-of-laws doctrines that weigh multiple connecting elements rather than relying on a single territorial fact (Leiter, Reference Leiter2020, p. 7).
6.3.2. Shared or concurrent jurisdiction
In cases involving cross-border AI systems, states could adopt frameworks for shared jurisdiction, similar to joint investigations under environmental or antitrust regimes. Co-regulation agreements would allow affected states to cooperate on supervision, audits, and enforcement while avoiding duplication. Such mechanisms already exist in the European Data Protection Board’s cross-border complaint procedures and could be replicated globally.
6.3.3. Digital commons and neutral zones
Certain domains, such as open-source AI models or shared research datasets, could be recognised as digital commons, governed by international standards rather than exclusive territorial claims. This idea parallels the regime of the high seas or outer space, where cooperative stewardship replaces unilateral sovereignty. UNESCO’s Recommendation on the Ethics of AI (2021) already points in this direction by urging open collaboration on AI for sustainable development.
6.3.4. International adjudicatory mechanisms
Finally, disputes arising from overlapping digital jurisdictions could be referred to specialised international panels or arbitral bodies. These could operate under the auspices of existing institutions such as the World Trade Organization or the International Court of Justice, or through a new Digital Governance Tribunal. Such mechanisms would lend predictability and legitimacy to jurisdictional determinations, reducing the risk of conflict.
Asian experiences offer distinctive insights for reconceptualising territoriality. China demonstrates how digital infrastructure can serve as a tool of sovereign assertion, embedding national ideology in code and architecture. India exemplifies functional sovereignty, using digital public infrastructure to mediate between openness and control (Srinath, Reference Srinath2024, p. 75). ASEAN, by contrast, highlights cooperative territoriality, where sovereignty is pooled through voluntary coordination rather than ceded to a supranational authority (Turobov, Reference Turobov2025). These variations suggest that digital territoriality need not be uniform; it can accommodate plural pathways reflecting historical, political, and developmental contexts.
In a global perspective, therefore, the reconfiguration of territoriality should not aim at erasing borders but at making them permeable and interoperable. Digital boundaries can coexist with global norms if grounded in transparency, reciprocity, and respect for human rights. As Avbelj (Reference Avbelj2024, p. 1092) argues, “territoriality in the algorithmic society must evolve from a metaphor of enclosure to one of coordination.” The ultimate objective is not to abolish sovereignty but to transform it into a relational concept that enables lawful interdependence.
7. Cross-border accountability mechanisms
As algorithms increasingly mediate transactions, communications, and governance across borders, the question of accountability has emerged as one of the central legal and ethical dilemmas of the digital age. Traditional doctrines of international responsibility are ill-equipped to address harms caused by autonomous or semi-autonomous systems that operate beyond a single jurisdiction. The difficulty lies not only in identifying the responsible actor, whether the developer, deployer, or state, but also in devising mechanisms through which injured parties can seek redress. This section examines how cross-border accountability might be reimagined through the lenses of state responsibility, transnational corporate liability, and cooperative enforcement mechanisms.
Under classical international law, the doctrine of state responsibility requires attribution of conduct to a state before legal liability can arise (International Law Commission, 2001, p. 34). Yet in the context of AI, attribution becomes complex. Algorithms are often created by private entities, trained using global datasets, and deployed through transnational infrastructures. When a government uses a privately developed algorithm for administrative decision-making, or when private platforms deploy AI systems with public consequences, the boundary between state and non-state action blurs (Crootof, Reference Crootof2018, p. 573).
For instance, predictive policing algorithms procured by public authorities but developed by private contractors can perpetuate discriminatory outcomes. The resulting harms, ranging from wrongful arrests to privacy violations, raise the question: who bears responsibility? If the algorithm’s bias results from flawed training data, does liability rest with the developer, the procuring state, or both? Similarly, when an autonomous trading algorithm triggers financial volatility across jurisdictions, determining the locus of control and intent is exceedingly difficult. These scenarios reveal what Fleur Johns (Reference Johns2023, p. 51) calls the “diffusion of agency in digital governance,” where accountability becomes dispersed among multiple human and non-human actors.
The International Law Commission’s Articles on State Responsibility provide a starting point, particularly Article 5, which attributes conduct to the state when a private entity exercises elements of governmental authority. This principle could extend to situations where states delegate decision-making functions to algorithmic systems. However, where private platforms operate independently but produce public harms, alternative routes such as due diligence obligations and extraterritorial human rights duties must be explored (McGregor, Reference McGregor2021, p. 212).
Recent jurisprudence in human rights law has expanded the concept of jurisdiction beyond strict territorial boundaries, recognising that states may have obligations where they exercise effective control over activities or individuals, even outside their territory. The European Court of Human Rights in Al-Skeini v. United Kingdom (2011) affirmed that the exercise of authority triggers jurisdictional responsibility. Applied to digital governance, this reasoning implies that when a state’s digital or algorithmic policies have extraterritorial effects, such as surveillance of foreign nationals or cross-border data profiling, it may still bear human rights obligations (Kuner, Reference Kuner2021, p. 478).
The United Nations Guiding Principles on Business and Human Rights (UNGPs) reinforce this logic by requiring states to ensure that companies domiciled within their territory respect human rights throughout global operations. This includes digital enterprises that deploy AI systems abroad. Some jurisdictions, such as France through its Duty of Vigilance Law (2017), have begun to operationalise these obligations, compelling multinational corporations to conduct due diligence on potential human rights impacts. Extending such frameworks to algorithmic systems would anchor accountability in existing human rights architecture.
Corporate actors play a pivotal role in shaping algorithmic ecosystems. However, the transnational nature of digital business complicates liability. Domestic laws are often inadequate to hold corporations accountable for harms occurring abroad, and international law traditionally addresses only states. The emerging concept of transnational corporate accountability seeks to bridge this gap by integrating principles from tort law, human rights, and trade regulation (Ruggie, Reference Ruggie2020, p. 319).
One promising avenue lies in the duty of care doctrine, which has been invoked in cross-border environmental and labour cases. In Vedanta Resources Plc v. Lungowe (2019), the UK Supreme Court recognised that a parent company could owe a duty of care for harms caused by its foreign subsidiary. By analogy, a technology corporation could bear responsibility for harms caused by algorithmic tools it develops or licenses abroad if it exercises substantial control over their design or deployment. This reasoning aligns with proposals for AI supply chain accountability, under which liability would extend through tiers of algorithmic production and use, similar to environmental and labour due diligence models (Yeung, Reference Yeung2023, p. 141).
Furthermore, recent initiatives such as the EU’s Corporate Sustainability Due Diligence Directive (2024) and the OECD AI Principles (2019) point towards a convergence of corporate and human rights accountability. They require risk assessment, transparency, and remediation mechanisms for adverse impacts arising from automated decision-making. For Asia, adopting similar standards could strengthen regional governance, ensuring that technology firms operating across borders respect local laws while adhering to global norms.
While doctrines of responsibility establish who is accountable, they do not necessarily ensure enforceability. Traditional mechanisms such as diplomatic protection or domestic litigation are often slow, fragmented, and inaccessible to affected individuals. To address algorithmic harms that transcend borders, international law requires new cooperative and hybrid enforcement mechanisms.
7.1. Cross-border investigative networks
Regional or international agencies could coordinate joint investigations into algorithmic incidents with transnational effects. Similar to existing models in environmental law and competition policy, these networks could share data, methodologies, and findings. For example, the Global Privacy Assembly (GPA) already facilitates coordination among national data protection authorities and could be expanded to include algorithmic impact assessments.
7.2. Algorithmic impact tribunals
Establishing a specialised Algorithmic Accountability Tribunal under UN auspices could provide a neutral forum for resolving disputes involving cross-border AI harms. Such a body would not replace domestic jurisdiction but complement it, functioning like the Permanent Court of Arbitration in the field of digital disputes. The tribunal could issue advisory opinions, recommend reparations, and promote harmonisation of standards across jurisdictions.
7.3. Soft law and peer review frameworks
In addition to formal adjudication, peer-review mechanisms similar to the Universal Periodic Review (UPR) process under the Human Rights Council could be used to assess states’ compliance with algorithmic accountability norms. These reviews would encourage transparency and foster a culture of cooperative compliance rather than punitive enforcement.
7.4. Transnational remedies for individuals
Finally, mechanisms enabling individuals to seek redress across borders through class actions, ombudsman offices, or grievance portals would democratise accountability. The Council of Europe’s AI Convention (2024) already recommends that signatories establish accessible remedies for algorithmic harms, setting a precedent for regional replication in Asia and beyond (Council of Europe, 2024).
Asia’s regional experience demonstrates that cooperative accountability is achievable even in politically diverse environments. ASEAN’s Digital Masterplan 2025 and its subsequent initiatives on cross-border data flows have laid the foundation for mutual recognition of digital standards (Turobov, Reference Turobov2025). Similarly, India and Japan have championed the Data Free Flow with Trust (DFFT) initiative at the G20, promoting international cooperation while safeguarding privacy and security. These efforts, though nascent, reflect a pragmatic approach to shared governance anchoring accountability in regional solidarity rather than rigid supranationalism.
The future of cross-border algorithmic accountability, therefore, lies in multi-layered governance. States must uphold their sovereign responsibilities, corporations must adhere to transnational duties, and international institutions must provide the connective tissue through which justice can be achieved. Such a system would not replace sovereignty but recalibrate it, transforming accountability from a zero-sum assertion of jurisdiction into a collective enterprise of global responsibility.
8. Treaty frameworks and cooperative regulation
Given the absence of a comprehensive global AI treaty, this section does not seek to catalogue all emerging initiatives but instead identifies representative instruments that illustrate the direction and limits of treaty-based cooperation. The rapid diffusion of AI technologies across borders exposes the limitations of fragmented national regimes. As algorithms shape global communication, trade, and governance, international law faces the challenge of developing cooperative regulatory frameworks that balance state sovereignty with collective responsibility. While unilateral or regional approaches can address specific concerns, only treaty-based cooperation can provide the predictability, reciprocity, and legitimacy required for sustainable global governance of AI.
The idea of using treaties to regulate technological risks is not new. Precedents exist in domains such as climate change, nuclear non-proliferation, and cybercrime, where states recognised that unilateral regulation was inadequate to address global externalities. Similarly, the algorithmic turn in global governance generates externalities that transcend jurisdictional lines bias in transnational hiring systems, misinformation propagated by recommendation algorithms, and cross-border privacy violations (Avbelj, Reference Avbelj2024, p. 1088).
Unlike earlier technological regimes, however, AI operates within privatised architectures controlled largely by non-state actors. Thus, treaty frameworks must accommodate both public and private governance dimensions. A well-designed international instrument could embed principles of transparency, accountability, and human oversight while allowing flexibility for domestic implementation. Fleming (Reference Fleming2025) argues that such treaties should act as “constitutional anchors for a fragmented digital order,” ensuring that states cooperate without relinquishing sovereignty.
Several international and regional initiatives already lay the groundwork for treaty-based AI governance:
8.1. The council of Europe’s framework convention on artificial intelligence (2024)
This convention represents the first binding international treaty explicitly addressing AI and human rights. It establishes obligations to ensure algorithmic transparency, non-discrimination, and effective remedies for harm. Importantly, it adopts a principles-based rather than a technology-specific approach, leaving room for national variation while reaffirming universal norms (Council of Europe, 2024). Although primarily European, the Convention is open to accession by non-member states, offering a template for broader global participation.
8.2. The OECD principles on AI (2019)
Endorsed by over 40 countries, these principles promote trustworthy AI grounded in human-centred values. While not legally binding, they provide a soft-law framework that informs national legislation and corporate policies (OECD, 2019). Many Asian jurisdictions, including Japan and South Korea, have integrated these principles into their domestic guidelines, demonstrating the feasibility of voluntary convergence.
8.3. The global partnership on artificial intelligence (GPAI)
Initiated by Canada and France and joined by several Asian members, GPAI functions as a multilateral forum for policy coordination and capacity building. It represents a hybrid model of public–private cooperation, engaging governments, academia, and industry in the co-production of ethical AI standards. Though non-treaty-based, GPAI’s networked structure could serve as an institutional foundation for future formal agreements (Hung, Reference Hung2025, p. 130).
Together, these initiatives illustrate the evolving architecture of multi-level governance, where binding treaties, soft-law instruments, and cooperative forums complement one another rather than compete.
Drawing lessons from these precedents, a global treaty on algorithmic governance should rest on four foundational pillars:
8.3.1. Principle of proportional sovereignty
Rather than treating sovereignty as absolute, the treaty should adopt a relational understanding recognising that the exercise of one state’s digital authority can affect the interests of others. The principle of proportional sovereignty would require states to regulate AI systems in a manner proportionate to their cross-border impacts (Leiter, Reference Leiter2020, p. 9).
8.3.2. Mutual recognition of algorithmic standards
States should commit to recognising each other’s certification or auditing mechanisms for AI systems, akin to mutual recognition agreements in trade law. This would reduce regulatory duplication and facilitate interoperability while maintaining minimum global standards (Avbelj, Reference Avbelj2024, p. 1091).
8.3.3. Shared responsibility and due diligence
The treaty could codify states’ obligations to exercise due diligence over AI systems developed or deployed within their jurisdiction, especially where these systems have foreseeable extraterritorial effects. This aligns with existing doctrines in environmental law and human rights, which hold states responsible for preventing transboundary harm (Kuner, Reference Kuner2021, p. 482).
8.3.4. Mechanisms for dispute resolution and peer review
An independent dispute-settlement body, modelled on the WTO’s Appellate Body or the Paris Agreement’s Compliance Mechanism, could adjudicate disputes concerning treaty implementation. Complementarily, a peer-review process similar to the Universal Periodic Review could assess compliance through dialogue and capacity building rather than coercion.
While a universal treaty remains aspirational, regional frameworks can serve as incubators of cooperation. Asia, in particular, offers fertile ground for developing treaty-like instruments through flexible and incremental methods.
ASEAN’s Digital Economy Framework Agreement (DEFA), expected to be finalised by 2025, illustrates how regional treaties can harmonise data protection and AI ethics across diverse legal systems (Turobov, Reference Turobov2025). By adopting a principle-based framework rather than prescriptive rules, DEFA preserves national sovereignty while ensuring compatibility across member states.
Similarly, the India–Japan Digital Partnership promotes cross-border data flows with trust, combining technical cooperation with normative alignment. Such bilateral or plurilateral initiatives may evolve into legally binding accords once mutual confidence and institutional capacity mature (Unnikrishnan, Reference Unnikrishnan2025).
Asia’s experience demonstrates that legal pluralism and treaty cooperation need not be contradictory. Instead, they can co-exist through flexible instruments, memoranda of understanding, soft-law commitments, and phased treaties that gradually deepen normative convergence.
Given the private sector’s dominance in AI development, any meaningful treaty framework must incorporate non-state actors not merely as stakeholders but as normative participants. Corporations, research institutions, and civil society groups should be engaged through consultative bodies similar to the UN Global Compact or Internet Governance Forum.
Including such actors can enhance legitimacy and compliance. As Hung (Reference Hung2025, p. 133) observes, “the governance of AI cannot succeed through intergovernmental diplomacy alone; it must harness the plural sites of authority that constitute the digital ecosystem.” Treaty mechanisms could thus require states to establish domestic multi-stakeholder councils to ensure continuous dialogue between regulators and innovators.
Sceptics argue that treaty-making is too slow for the fast-paced evolution of AI technology. Negotiating consensus among states with divergent political systems and technological capacities is undeniably difficult. Yet, as history shows, multilateralism often follows initial fragmentation. The Paris Agreement (2015) emerged after decades of disjointed environmental treaties, and the Budapest Convention on Cybercrime (2001) achieved broad adherence despite initial scepticism.
Moreover, the value of treaties lies not only in enforceability but also in normative signalling, articulating shared expectations that guide behaviour even before formal ratification. As Avbelj (Reference Avbelj2024, p. 1092) notes, “treaty frameworks function as moral compasses in an age of legal dissonance.” For AI governance, such compasses are essential to counterbalance technological determinism with legal and ethical restraint.
9. Conclusion and policy implications
The rise of algorithmic governance represents one of the most significant transformations in the history of law’s engagement with technology. AI has reconfigured how authority is exercised, how accountability is distributed, and how sovereignty is conceived. Yet, as this article has argued, international law is not powerless in the face of these shifts. Rather, it must adapt its foundational concepts of sovereignty, territoriality, and responsibility to the networked and interdependent realities of the digital age.
At its core, this study has advanced the claim that AI sovereignty does not necessarily threaten the international legal order but instead provides an opportunity to reimagine it. States are no longer the sole bearers of authority; they now share governance with private actors, algorithms, and transnational institutions. This pluralisation of authority requires equally plural mechanisms of accountability, grounded not in rigid territorial control but in cooperation, transparency, and shared responsibility.
The first major insight concerns the evolution of sovereignty itself. The transition from data sovereignty to algorithmic sovereignty signifies a movement from controlling information to shaping the logic of decision-making systems. Asian jurisdictions, particularly China, India, and ASEAN, illustrate that sovereignty today is not a static possession but a dynamic capability to define normative boundaries within global digital networks (Hung, Reference Hung2025, p. 119; Srinath, Reference Srinath2024, p. 70).
However, the assertion of digital sovereignty cannot be sustained through isolation. The reconceptualisation of territoriality proposed in this article suggests that sovereignty must be reinterpreted as a relational construct exercised through coordination rather than exclusion. States can no longer treat cyberspace as an extension of physical territory; they must engage in functional and cooperative jurisdiction, aligning domestic authority with transnational impacts. This approach preserves sovereign agency while recognising the inherently global character of AI infrastructures.
The second insight is that algorithmic accountability requires new legal architectures of responsibility. Traditional doctrines of state responsibility, built upon attribution and control, are insufficient for autonomous or semi-autonomous systems whose actions are often emergent and opaque (Crootof, Reference Crootof2018, p. 574). Accountability, therefore, must extend across the entire lifecycle of AI systems from design to deployment, encompassing states, corporations, and individuals.
Mechanisms such as transnational due diligence obligations, shared investigative networks, and peer-review frameworks can make accountability practical rather than rhetorical. Asian initiatives like ASEAN’s Digital Masterplan 2025 and India–Japan’s Data Free Flow with Trust partnership exemplify how soft-law cooperation can evolve into structured regimes of mutual oversight (Turobov, Reference Turobov2025; Unnikrishnan, Reference Unnikrishnan2025). By embedding procedural fairness, human rights, and transparency into algorithmic governance, these frameworks operationalise the moral commitments that underlie international law.
The third insight highlights the necessity of treaty-based cooperation. Fragmented national laws and voluntary guidelines cannot manage the scale and complexity of AI’s global footprint. A future Global AI Treaty grounded in principles of proportional sovereignty, mutual recognition, and due diligence could institutionalise accountability without dismantling state autonomy. Precedents such as the Council of Europe’s AI Convention (2024), the OECD Principles on AI (2019), and the GPAI framework already demonstrate the feasibility of incremental multilateralism (Council of Europe, 2024; OECD, 2019; Hung, Reference Hung2025, p. 22).
Asia’s regional pragmatism offers valuable lessons for this endeavour. The ASEAN model of consensual integration and India’s rights-based digital policy show that international cooperation need not follow a single path. Rather, it can evolve through adaptive pluralism, where regional treaties, bilateral accords, and soft-law mechanisms collectively constitute a layered regime of algorithmic governance.
The implications of this analysis extend beyond academic theory into concrete policymaking.
States should embed algorithmic due diligence obligations in administrative and corporate law, ensuring that all public and private uses of AI conform to constitutional and human rights principles. Independent audit mechanisms, public registers of high-risk algorithms, and accessible grievance redressal systems should form part of domestic governance frameworks.
Regional organisations, especially ASEAN, SAARC, and APEC, should establish Digital Ethics Councils to promote interoperability and data trust. These bodies can facilitate information sharing, joint investigations, and capacity building across member states, reducing regulatory asymmetry.
Multilateral institutions must adopt a more inclusive approach to AI governance by engaging non-state actors as co-regulators. A UN Digital Governance Forum could harmonise existing efforts, convening states, corporations, and civil society to negotiate standards. Moreover, the creation of an Algorithmic Accountability Tribunal could institutionalise remedies for cross-border digital harms. Companies developing or deploying AI systems should adopt transparent documentation and public impact assessments, while civil society organisations should act as watchdogs to ensure democratic oversight. Ethical certification labels similar to environmental sustainability seals could signal compliance with human rights and fairness standards.
The broader ambition of this article is to reposition international law as an enabling framework for the digital future. Rather than resisting technological change, international law must become an instrument for steering it towards legitimacy and justice. Legal pluralism acknowledges diversity; global constitutionalism anchors it in shared values. Together, they form the twin pillars of a reimagined legal order, one that reconciles sovereignty with solidarity.
As Avbelj (Reference Avbelj2024, p. 1092) aptly notes, “international law’s relevance lies not in its power to constrain but in its capacity to coordinate.” The coordination of algorithmic governance, though complex, is achievable through layered cooperation, normative humility, and ethical foresight. Asia’s diverse legal traditions and policy innovations demonstrate that pluralism and principle can coexist and that sovereignty, when exercised responsibly, can sustain rather than subvert international law.
In sum, algorithmic governance forces the law to confront its most fundamental question: how can order be maintained in a world where power is distributed across humans and machines, states and networks? The answer lies not in reclaiming lost control but in designing lawful interdependence. Through reconceptualised territoriality, shared accountability, and treaty-based cooperation, international law can remain not only relevant but indispensable in the algorithmic century.
Open access
