Artificial Intelligence, Arbitration and the Global South: Addressing Algorithmic Asymmetries in Global Dispute Resolution(*)(**)

Inteligencia artificial, arbitraje y el Sur Global: abordando las asimetrías algorítmicas en la resolución internacional de controversias

Daniel Brantes Ferreira(***)

Founding Partner at DBFLaw (Rio de Janeiro, Brazil)

Elizaveta A. Gromova(****)

National Research South Ural State University (Chelyabinsk, Russia)

Abstract: This article examines whether the increasing use of artificial intelligence (AI) in international arbitration risks reinforcing and expanding Global North dominance over the Global South. While existing literature mainly focuses on efficiency, confidentiality, transparency, and due process in AI-assisted arbitration, this article addresses this gap by viewing AI not just as a technological innovation but as a form of private transnational governance embedded within broader structures of epistemic hierarchy, technological dependency, and Global North–South inequality. Methodologically, it combines doctrinal and literature-based legal analysis with a TWAIL-informed political economy perspective, drawing on scholarly literature, institutional reports, surveys, and soft-law instruments. The analysis highlights three main mechanisms through which inequalities may be reinforced: data concentration, vendor control, and the dominance of Global North standard-setting. It argues that AI may strengthen existing institutional, infrastructural, and normative advantages of actors at the center of the arbitral system. At the same time, the article does not oppose AI in arbitration. Instead, it concludes that AI should be governed with safeguards such as disclosure, explainability, contestability, confidentiality, and human oversight, while also addressing deeper inequalities in technological infrastructure and epistemic authority.

Keywords: Artificial Intelligence - International Arbitration - Global South – TWAIL - Data Colonialism - Vendor Control - Procedural Integrity - AI governance

1. Introduction

In the international arena, arbitration has long been the default method for resolving commercial and investment disputes. It is valued for efficiency, neutrality, finality, enforceability, and confidentiality. International agreements like the 1958 New York Convention and the 1965 International Convention on the Settlement of Investment Disputes support its use. The New York Convention alone has 172 signatory members (United Nations Commission on International Trade Law, n.d.). However, critics argue that arbitration is not always cost-effective or quick (Gicquello, 2020). Sometimes it is costlier and slower than court litigation. These delays often happen due to heavy provider backlogs or party actions (Najar, 2009). Dispute resolution providers are innovating with new frameworks to keep arbitration attractive.

A 2025 study by White & Case LLP and the School of International Arbitration at Queen Mary University of London found parties most often choose expedited procedures to lower costs (65%) and speed up proceedings (58%). Technology is also key to maintaining arbitration’s appeal. The same study found that the main reasons parties adopt AI are to save time (54%), reduce costs (44%), and decrease human error (39%). Some behaviors, such as lawyers’ adversarial tactics (24%) and arbitrators’ inadequate case management (23%), harm the process. The survey shows that institutions are adopting new rules to boost efficiency. Innovations include expedited arbitration, early decisions, consolidation, joinder, and emergency measures. Arbitration now depends on both these rules and technological progress. Technology and AI are now central to arbitration.

The focus of this analysis moves beyond technological tools and cost-benefit considerations to examine governance, asserting that regulatory and technological frameworks are central in shaping transnational disputes. When such frameworks exist, they structure disputes; in their absence, resolution becomes more complex. Transnational private governance is identified as essential to both arbitration and AI development, particularly in rulemaking, application, and enforcement (Whytock, 2008). Governance influences each domain and its integration in distinct ways. Unlike judicial procedure, arbitration is largely structured through privately governed institutional rules developed outside public legislative processes, often within competitive service-provider environments that may prioritize efficiency, flexibility, and market responsiveness (Horton, 2020). This creates a parallel procedural system, a phenomenon also observed in AI governance. Commercial arbitration exemplifies private adjudication (Landes & Posner, 1979), offering market-driven judicial services shaped by institutional competition among providers, law firms, and arbitrators. The arbitration market remains concentrated in the Global North, both economically and geopolitically.

Building on this analysis of transnational private governance, states now position themselves as attractive, arbitration-friendly jurisdictions. Many have enacted or revised laws to align with the 1985 UNCITRAL Model Law (Drahozal, 2004), while others have modernized laws for new technologies and institutional innovations. Currently, 93 states and 127 jurisdictions have adopted or been influenced by the Model Law (United Nations Commission on International Trade Law, n.d.). Since 2020, 13 states and one jurisdiction have adopted such statutes, including Tonga (2020), Uzbekistan (2021), Sierra Leone (2022), Albania (2023), Greece (2023), Japan (2023), Nigeria (2023), Trinidad and Tobago (2023), Azerbaijan (2024), Costa Rica (2024), Guyana (2024), Israel (2024), Malawi (2024), and Puerto Rico (2024). Of these, only four, namely Greece, Japan, Israel, and Puerto Rico, are in the Global North; the remainder are classified as Global South jurisdictions. This trend highlights these countries’ drive to become leading arbitration venues, as legislative changes also function as marketing strategies (Berger, 1993), reinforcing each jurisdiction’s user-focused arbitration ecosystem.

Consequently, as technology continues to shape arbitration, integrating artificial intelligence into forums has procedural implications beyond mere technical assistance. AI tools currently serve many functions in arbitration (Solhchi & Baghbanno, 2023; Yıldırım, 2024). These include document review, predictive analysis, case management, and drafting support. The American Arbitration Association (AAA) recently launched an AI arbitrator for high-volume, low-complexity cases, under human oversight (American Arbitration Association, n.d.).

This AI framework shifts informational and strategic advantages. It enables outcome prediction and informs decisions on arbitrator selection, procedural tactics, and settlement timing. Entities with access to advanced data and proprietary models, primarily law firms, institutions, or companies that own sophisticated AI, may maintain private legal databases, predictive tools, and models trained specifically for arbitration. A select group of organizations predominantly controls these expensive resources. This group includes major Global North tech firms. Use of AI in arbitration raises several concerns, including reliance on data, potential confidentiality breaches, limited training data, and machine bias and opacity (a lack of transparency in algorithmic decisions) (Kasap, 2021). These challenges are significant because AI outputs can influence assessment and reasoning in disputes.

Regulatory questions are central. Emerging frameworks and guidelines from arbitration institutions and professional bodies are critical (Tsuvina & Tsuvina, 2024). Much of the regulatory infrastructure originates from the Global North, such as the Chartered Institute of Arbitrators (CIArb) (Chartered Institute of Arbitrators, 2025) and the Silicon Valley Arbitration & Mediation Center (SVAMC) (Silicon Valley Arbitration & Mediation Center, 2024). This prevalence may reinforce inequities in global dispute resolution. The majority of AI tools and guidelines stem from the Global North. The only prominent Global South country advancing in this field is China. The China International Economic and Trade Arbitration Commission (CIETAC) has issued AI Guidelines (China International Economic and Trade Arbitration Commission, 2025), and the Guangzhou Arbitration Commission (GZAC) is at the forefront of AI use in arbitration.

Given the challenges from technological and regulatory shifts, the use of technology in arbitration calls for basic procedural safeguards. For example, due process and the right to present a case safeguard procedural integrity while supporting innovation. If institutions require explainable outcomes, algorithmic decision-making can advance both arbitration and fairness (Fortes, 2020; Pasquale, 2017). In this context, algorithmic transparency becomes crucial. Without it, whether in data collection, code, or system development, companies and law firms may gain unfair advantages. Auditing application code is one solution that addresses this concern (Susskind, 2018).

In this scenario, AI rules in arbitration are not neutral. Powerful institutions, especially those in the Global North, establish both AI guidelines and the technology behind arbitration standards, shaping governance from a Foucauldian perspective (Foucault, 2013). Similarly, the political economy approach views legal technologies and regulatory choices as rooted in historical power structures rather than as neutral tools (Cox, 1981). Regulatory frameworks for AI in arbitration are therefore expressions of broader epistemic and institutional power, not merely technical fixes. As power is closely tied to economic dominance, it is unsurprising that most AI guidelines originate from the Global North’s legal and technological environments.

In light of these arguments, the central question is whether the proliferation of AI in international arbitration risks reinforcing and expanding the Global North’s dominance over the Global South. Three mechanisms are identified for analysis: data concentration, vendor control, and the adoption of governance standards originating from the Global North. The key hypothesis is that these dynamics may not only enhance efficiency but also exacerbate existing asymmetries in arbitration. Ultimately, AI has the potential to extend the epistemic and infrastructural advantages of arbitration providers and users in the Global North, particularly regarding preferred venues, standards, institutions, and datasets.

Although recent studies address transparency, confidentiality, and due process in AI-assisted arbitration, the literature still lacks a thorough analysis of how data concentration, vendor control, and Global North standard-setting may reinforce structural inequalities between the Global North and the Global South. This article fills that gap by examining AI in arbitration not merely as a procedural innovation, but as a form of transnational private governance embedded in broader dynamics of epistemic hierarchy, technological dependency, and North–South inequality.

Methodologically, this article combines doctrinal and literature-based legal analysis with a TWAIL-informed political economy perspective. It examines scholarly literature, institutional reports, surveys, and soft-law instruments to assess how AI governance in arbitration is shaped by asymmetries in data access, technological infrastructure, and normative production. The article also develops a normative framework for evaluating procedural safeguards in AI-assisted arbitration from a Global North–Global South perspective. The analysis primarily reviews scholarly literature, supplemented by grey literature, including institutional reports, surveys, and guidelines.

The subsequent sections are organized to align with the outlined methodology. Section 2 presents the theoretical and structural framework by examining the Global North–Global South divide in international arbitration, the concentration of arbitral ecosystems, and the interplay between private adjudication, market dynamics, and emerging forms of digital and epistemic domination. Section 3 analyzes the integration of artificial intelligence into arbitration, focusing on its technological roles, governance frameworks, structural asymmetries, and the procedural safeguards needed to uphold due process, transparency, and contestability in AI-assisted proceedings. The article concludes by considering the broader implications of AI-driven arbitration for the future distribution of power in global dispute resolution.

2. The Global North–Global South Divide in International Arbitration

This section establishes the theoretical framework for the subsequent analysis. It argues that the Global North–Global South divide in international arbitration must be understood within the larger historical and structural inequalities of the international legal order. TWAIL offers the most relevant critical perspective for this purpose, as it uncovers how domination is maintained not only through formal legal rules and institutions but also via governance structures, epistemic hierarchies, and technological infrastructures.

2.1. TWAIL Beyond Classical International Law: Technology, Arbitration, and New Forms of Global Hierarchy

Third World Approaches to International Law (TWAIL) is best understood as a critical intellectual and political movement that aims to reveal how international law has historically contributed to the subjugation of the Third World and continues to uphold global hierarchy today. In a classic formulation, Mutua (2000) describes TWAIL as a broad oppositional project focused on how international law legitimizes and maintains the unequal structures of the international order. Similarly, Chimni (2006) states that international law has become “the principal language in which domination is coming to be expressed in the era of globalization,” as it helps legitimize and sustain the growing North–South divide. TWAIL, therefore, is not just a doctrinal critique of specific rules but a broader methodological and normative effort to examine the relationship among law, empire, capitalism, and resistance (Okafor, 2008).

In this sense, TWAIL scholarship should be understood not merely as a body of doctrine but as a critical methodological framework that interrogates the historical and structural conditions under which international law operates. As Rajagopal (2003) emphasizes, TWAIL scholarship also seeks to recover the agency of Third World actors and social movements in shaping international legal norms and institutions.

The intellectual history of TWAIL is often explained through a distinction between TWAIL I and TWAIL II, although that periodization should not be viewed as fixed. TWAIL I is generally associated with the anti-colonial and decolonization era, when jurists from Asia, Africa, and the broader Third World challenged the Eurocentric assumptions of classical international law and called for a more universal international legal order (Anand, 1962; 1966a; 1966b; 1967). In this earlier phase, the focus was on sovereignty, sovereign equality, self-determination, non-intervention, and restructuring international economic relations. TWAIL II, on the other hand, is typically associated with a later group of scholars who moved beyond external critique of exclusion to argue that colonialism was embedded in international law itself. Anghie’s work (1996) on the colonial roots of international law, Mutua’s critique of international law as an imperial structure, and Chimni’s analysis of contemporary international institutions as part of a developing imperial global order all exemplify this second stage. As Chimni (2006) describes, the relationship between state sovereignty and international law has been reshaped in the era of globalization, weakening the autonomy of Third World states and peoples and shifting economic power to international institutions and transnational actors (Rajagopal, 2003).

At the same time, the TWAIL I/TWAIL II distinction should be approached carefully. Gathii (2011) emphasizes that TWAIL is a decentralized and evolving intellectual network rather than a fixed canon, while Galindo (2016) warns that strict periodization might obscure continuities within Third World critiques of international law. Although this article does not claim the existence of a fully stabilized “TWAIL III,” it asserts that an emerging third TWAIL phase can be identified in the governance of technology and arbitration. At this point, Boaventura de Sousa Santos becomes especially significant. His critique of abyssal thinking explains how modern Western knowledge draws visible and invisible lines that make alternative knowledges nonexistent, subordinated, or irrelevant (Santos, 2007). Similarly, his sociology of absences and emergences demonstrates how hegemonic structures waste social experiences and suppress counter-hegemonic forms of knowledge production (Santos, 2002). When applied to arbitration and artificial intelligence, this indicates that Global North institutions do not only materially dominate rulemaking; they also influence which datasets, standards, and expertise are considered authoritative. In this light, the interaction between AI, arbitration, and private transnational governance reflects an emerging third TWAIL moment focused on technological control, arbitral governance, and epistemic authority.

The broader critical literature on postcolonial power and the universalizing tendencies of international law supports this argument. Pahuja’s analysis (Pahuja, 2011) is especially significant in this context. As she demonstrates, the failure of international law in the postcolonial world is linked to the universalization of development and economic growth as regulatory rationalities, transforming international law into a form of power rather than a neutral framework for emancipation. The outcome is a technical, universal legal order that claims neutrality while maintaining asymmetric relations with the North.

Chimni’s manifesto (2006) emphasizes that globalization has reshaped the relationship between the state and international law, benefiting transnational capital and diminishing the autonomy of the Third World. This is evident in international institutions, the internationalization of property rights, and the spread of transnational adjudicatory mechanisms such as international commercial arbitration. In this context, arbitration is not separate from the TWAIL issue; it is a key platform through which global hierarchy is reinforced.

From this perspective, a third TWAIL moment can be seen as emerging around the Global North’s dominance not only over international law in the traditional sense but also over the infrastructures that now shape international commerce and dispute resolution. Artificial intelligence, proprietary datasets, digital platforms, arbitral institutions, soft law guidelines, and professional expertise are increasingly concentrated within Global North legal and technological systems. This concentration enables the North to influence not just the substance of transnational dispute resolution but also the epistemic and technological conditions under which disputes are processed, interpreted, and governed. Therefore, the issue is no longer limited to the colonial roots of international law or the imperial nature of international institutions. It also involves the expansion of Global North dominance through the governance of arbitration and technology, and the merging of both in AI-assisted dispute resolution. In this way, this article argues that the interaction between artificial intelligence, arbitration, and private transnational governance may reflect an emerging third TWAIL moment: one centered on how technological control, arbitral governance, and epistemic authority together perpetuate new forms of North–South subordination in international economic life.

2.2. Arbitration as Private Adjudication, Market and Geographic Concentration of the Arbitral Ecosystems

International arbitration is best characterized as private adjudication within a competitive, transnational marketplace for dispute resolution. As Landes and Posner (1979) note, adjudication is not inherently public; commercial arbitration exemplifies its private form. In this market, parties deliberately select rules, arbitrators, and the institutional and territorial frameworks for their disputes. Arbitral institutions, seats, professional service providers, and national legal systems, in turn, compete for users by offering distinct reputational, procedural, and legal advantages (Landes & Posner, 1979; Drahozal, 2004). Thus, arbitration’s distinctive feature is not merely that it is private, but that it operates within a global market for adjudicatory services.

This market-oriented character helps explain why arbitration has developed around a small number of highly reputed arbitral ecosystems. Surveys by Queen Mary University of London and White & Case show that users continue to prefer a limited group of seats and institutions. In 2015, the five most preferred seats were London, Paris, Hong Kong, Singapore, and Geneva. The five top institutions were the International Chamber of Commerce (ICC), the London Court of International Arbitration (LCIA), the Hong Kong International Arbitration Centre (HKIAC), the Singapore International Arbitration Centre (SIAC), and the Stockholm Chamber of Commerce Arbitration Institute (SCC) (White & Case LLP & School of International Arbitration, Queen Mary University of London, 2015). The 2018 survey identified London, Paris, Singapore, Hong Kong, and Geneva as the leading seats. ICC, LCIA, SIAC, HKIAC, and SCC again ranked as the top institutions. Similarly, the 2021 survey reported London, Singapore, Hong Kong, Paris, and Geneva as the top seats. The top institutions were ICC, SIAC, HKIAC, LCIA, and CIETAC. In 2025, the five most preferred seats shifted to London, Singapore, Hong Kong, Beijing, and Paris. The preferred rules were ICC, HKIAC, SIAC, LCIA, and UNCITRAL. These findings reveal both persistence and change: the market remains concentrated, but regional arbitral centers, especially in Asia, have achieved prominence.

The persistence of London, Paris, Geneva, Singapore, and Hong Kong across successive surveys demonstrates the stickiness of established arbitral hubs. At the same time, Beijing’s rise in the 2025 survey and the growing visibility of Chinese institutions such as CIETAC in the 2021 survey suggest that the geography of arbitration is evolving rather than fixed. This evolution, however, should not be mistaken for a full decentralization of arbitral power. Rather, it indicates a selective expansion of leading regional centers within a still highly concentrated global market. In other words, new participants may enter the upper tiers of arbitral competition. However, they do so in a field whose structure is still organized around prestige, enforceability, institutional experience, and legal infrastructure. These features have historically favored Global North seats and, more recently, a limited number of powerful Asian hubs, such as Singapore, have a dual identity: historically and geographically classified as part of the Global South due to its colonial past, but operating as high-income technology hubs, which is typical of the Global North (The GeoEconomics, n.d.).

This pattern fits the logic of regulatory competition discussed by Drahozal (2004). Parties often choose the place of arbitration with close attention to the seat’s legal environment. As a result, states have strong incentives to modernize arbitration laws and signal user-friendliness to the international arbitration community. Drahozal (2004) shows that when countries revise their arbitration statutes, they may attract more arbitration proceedings. This is true even if the economic effects are more modest than reformers hope. The legal environment becomes part of the market. National arbitration laws, judicial support, enforceability records, and reputational signaling all serve as competitive assets in attracting arbitral business. This context helps explain why the diffusion of Model Law-based legislation, the growth of modern arbitral institutions, and the appearance of new regional seats are not isolated reforms. Instead, they are part of a broader market for transnational adjudicatory services.

From this perspective, the increasing prominence of regional arbitral centers, particularly in China, does not displace the argument about structural concentration; rather, it refines it. What emerges from the empirical record is not a flat, fully plural arbitral order, but a hierarchically organized ecosystem in which a small number of seats and institutions persistently dominate user preferences, even as selected regional actors, such as Beijing, Guangzhou, Shenzhen, and Dubai, gain influence. International arbitration, therefore, remains a market of private adjudication, characterized by both competition and concentration, and its geography continues to display a broader distribution of economic power, legal prestige, and institutional authority.

The market-driven aspect of international arbitration also encourages forum shopping, the strategic choice of the most advantageous forum from among available procedural, institutional, and territorial options. In transnational disputes, forum shopping should not be viewed merely as opportunistic manipulation; rather, it is often a rational response to variations in costs, predictability, legal infrastructure, and enforcement prospects across different forums (Assareh, 2013; Bookman, 2016). Parties are frequently able to assess key variables in advance, such as forum fees, adjudicator fees, and other access-to-justice costs, making the selection of a forum an essential part of dispute resolution planning. Likewise, global forum shopping can serve important functions, including safeguarding access to justice, promoting private enforcement, and even fostering legal reform, rather than simply being an unfair tactic. It also involves strategic considerations about where precedent, remedies, and procedural leverage will be most valuable in future disputes (Busch, 2007).

In arbitration, this tendency often concentrates disputes within a few highly regarded seats and institutions, as parties typically favor forums that offer legal certainty, procedural sophistication, enforceability, strong reputations, and sit in arbitration-friendly jurisdictions. At the same time, as new regional centers seek to attract disputes, arbitral forum shopping increasingly serves as both a sign and a driver of competition among arbitral ecosystems, including emerging investment corridors and treaty-based dispute-resolution frameworks (Idrees et al., 2019). An example of an emerging investment corridor is the China–Pakistan Economic Corridor (CPEC), which has created new pressures for arbitral forum selection and dispute-resolution planning related to large-scale infrastructure, energy, and connectivity projects (Idrees et al., 2019; McLaughlin, 2023).

Empirical data from leading arbitral institutions further confirm the concentration of international arbitration within a few highly reputable institutional hubs, while also highlighting the growing participation of parties from the Global South. For analysis, this section focuses on the International Chamber of Commerce (ICC), the London Court of International Arbitration (LCIA), the Singapore International Arbitration Centre (SIAC), and the International Centre for Settlement of Investment Disputes (ICSID). These institutions were chosen because they are among the most influential dispute-resolution providers in the global arbitration market and collectively demonstrate patterns across commercial and investment arbitration. Their large caseloads, longstanding reputations, and key roles in shaping arbitral rules, best practices, and procedural innovations position them as gatekeepers of international arbitration governance. The analysis uses 2024 as the baseline year to ensure consistency in institutional statistics.

The LCIA continues to handle hundreds of international disputes annually, reaffirming its position as a leading Western arbitration institution (London Court of International Arbitration [LCIA], 2025). Likewise, SIAC reported a significant caseload in its 2024 annual report, with the majority of cases involving international disputes (Singapore International Arbitration Centre [SIAC], 2024). The ICC Court of Arbitration also manages several hundred new cases each year involving parties from numerous jurisdictions globally (International Chamber of Commerce [ICC], 2024). In investment arbitration, ICSID 2024 statistics also demonstrate the large and geographically diverse caseload administered under the ICSID Convention and Additional Facility Rules, underscoring ICSID’s central role in disputes involving states and investors from emerging economies (International Centre for Settlement of Investment Disputes [ICSID], 2024).

Although these institutions handle disputes involving parties from around the world, they remain embedded within institutional ecosystems historically linked to the Global North. London and Paris have long served as central hubs for international arbitration, while Singapore, despite its geographic location in Asia and its historical classification as part of the Global South, functions today as a high-income legal and financial center closely connected to Global North legal markets. Consequently, ICC, LCIA, SIAC, and ICSID form a transnational core of international arbitration, combining high caseloads, advanced legal infrastructure, and strong reputational capital. Their influence extends beyond case administration: these institutions actively shape arbitration governance through rule-making, institutional policies, conferences, and thought-leadership initiatives. In this way, they serve not only as service providers but also as norm-setters within the global arbitration ecosystem.

It should be noted that this institutional asymmetry does not imply that arbitration institutions make no effort to promote inclusiveness. On the contrary, many institutions seek to incorporate professionals from diverse jurisdictions worldwide, both as potential arbitrators and as participants in the institutional governance of arbitration providers (International Chamber of Commerce, n.d.). A notable example is the International Court of Arbitration of the International Chamber of Commerce (ICC), which currently comprises 170 members, of whom 69 are from Global North jurisdictions and 101 from Global South jurisdictions. Accordingly, representatives from the Global South account for approximately 59.4% of the Court’s composition. While this distribution may be seen as evidence of the institution’s commitment to geographical inclusiveness, it also reflects a strategic effort to ensure representation across the markets in which the institution operates, thereby reinforcing its global reach and institutional legitimacy.

At the same time, the statistics reported by these institutions reveal a notable structural imbalance in the arbitral marketplace. Although arbitration users have become geographically diverse, the institutional authority regulating dispute resolution remains highly centralized. Parties from the Global South frequently act as claimants, respondents, or contractual partners in complex cross-border disputes, particularly in sectors such as infrastructure, construction, energy, and international investment. However, the administration of these disputes often remains within a small group of globally recognized arbitral institutions such as ICC, LCIA, SIAC, and ICSID. As a result, international arbitration can be seen as globally diverse in participation but not fully decentralized in governance. The system increasingly includes Global South actors as users, but the institutional frameworks shaping arbitral procedures, expertise, and technological infrastructure remain concentrated in a few powerful arbitration hubs.

These patterns of arbitral concentration must also be considered alongside the expanding literature on digital and data colonialism. This research highlights how control over technological infrastructure, data extraction, and platform governance may perpetuate older North–South hierarchies in new forms. Couldry and Mejias (2018) argue that contemporary capitalism increasingly relies on the appropriation of human life through data, creating a new colonial logic where social activity itself becomes a resource for extraction. Similarly, Mann and Daly (2019) and Ferreira and Gromova (2025) show how informational power can serve as a form of digital colonialism, reinforcing asymmetries between the Global North and Global South through control over data flows, surveillance capabilities, and technological systems, and how humans are being quantified into data in their everyday lives. Schneider (2022) also highlights those digital infrastructures are not neutral but structured through “stacks” of governance and technical control that may deepen dependency when held by a small number of powerful actors.

In the arbitral context, this indicates that structural concentration is no longer solely geographic or institutional; it is also digital and epistemic. As arbitration increasingly depends on data-driven tools, proprietary platforms, and AI-assisted processes, the dominance of Global North institutions extends beyond seats and rules to the very infrastructures through which disputes are processed, interpreted, and governed. Therefore, the rise of AI-assisted arbitration occurs not on neutral ground but within a landscape already marked by asymmetries in infrastructure, data access, and technological power.

The concept of digital colonialism can be seen as a contrast to the ideas of digital and data sovereignty. Digital sovereignty refers to a country’s ability to control incoming information, strengthen its authority over online accounts, data, and the personally identifiable information (PII) of its residents, and increase trade protectionism in the digital sector (Grimm, 2015). In contrast, data sovereignty involves regulating data flow through filtering technologies on the internet and establishing rules and infrastructure to protect the country and its citizens, such as data storage and data protection (Karlstad, 2023; Gromova & Ferreira, 2024). Given these definitions, countries in the Global South clearly lack both digital and data sovereignty in the field of artificial intelligence and in the use of technological tools in arbitration, especially in the latter.

Seen in this broader context, the idea of an emerging third TWAIL moment becomes even more convincing when considered alongside the literature on digital and data colonialism. TWAIL I revealed the exclusion of the Third World from shaping classical international law, and TWAIL II showed how colonial dominance was embedded in international legal doctrines and institutions. A contemporary TWAIL III can be understood as addressing the expansion of those hierarchies into digital infrastructures, data collection, and technological governance. In the context of arbitration, this means that the focus on seats, institutions, and expertise is increasingly connected to data, platforms, and AI tools. Therefore, an emerging TWAIL III does not replace earlier critiques but builds on them: from colonialism and international legal doctrine to today’s regime, where arbitral governance and digital infrastructures together foster new forms of North–South subordination.

3. Artificial Intelligence in Arbitration: Governance, Asymmetries, and Procedural Integrity

Given this structural context, the question is no longer whether artificial intelligence will impact international arbitration, but rather how that influence will be structured, who will manage it, and what the distribution of benefits and burdens will be. Recent research shows that AI is already integrated into arbitral practices, including document review, legal research, predictive analytics, case management, and drafting assistance. This indicates that AI’s significance extends beyond procedural ease into the governance of dispute resolution itself (Gicquello, 2020; Tsuvina & Tsuvina, 2024).

As AI becomes increasingly embedded in the creation and sharing of arbitral knowledge, concerns about legitimacy, fairness, and due process must be addressed alongside issues of technological dependence and epistemic control (Fortes, 2020). In international arbitration, where institutional authority and professional influence are already unevenly distributed, the widespread adoption of AI risks entrenching current hierarchies unless its governance is critically examined and meaningful safeguards are put in place to protect its application. Therefore, it is essential to explore, first, how AI governance might reproduce existing structural inequalities within the arbitral field, and second, what procedural protections are necessary to uphold the integrity of AI-supported proceedings. In this sense, AI does not enter arbitration on neutral institutional terrain, but into a field already shaped by concentrated arbitral authority, market dependency, and epistemic hierarchy.

3.1. AI Governance and Structural Asymmetries in International Arbitration

Artificial intelligence in arbitration should not be understood merely as a set of efficiency-enhancing tools, but as an emerging layer of private governance that may reinforce existing asymmetries in data access, technological control, and normative authority within global dispute resolution. Recent scholarship has shown that AI can already perform, or assist with, a wide range of arbitral functions, including document review, legal research, predictive analytics, drafting support, and case management (Gicquello, 2020; Solhchi & Baghbanno, 2023; Yıldırım, 2024). Gicquello (2020) argues that the introduction of AI into international arbitration is often justified on the basis that it may enhance legitimacy, fairness, and efficiency while also engaging broader systemic concerns. At the same time, Tsuvina and Tsuvina (2024) correctly note that the growing use of large language models in arbitration raises not only practical questions of utility, but also regulatory and theoretical questions concerning how adjudicative processes are structured and governed. Once AI becomes part of how disputes are organized, assessed, and strategically managed, its significance is no longer purely technical. It becomes institutional, distributive, and ultimately political.

The structural risks examined in this section operate through three interrelated mechanisms: the unequal distribution of arbitral data, the growing control of privately owned technological infrastructures, and the concentration of AI-related rulemaking in Global North institutions and professional networks.

The first major asymmetry concerns data. AI systems depend on the availability, quality, and structure of datasets, yet arbitral data is unevenly distributed and often privately held. Unlike many domestic judicial systems, arbitration is marked by confidentiality, fragmented publication practices, and unequal access to awards, pleadings, procedural orders, and case management records. Łągiewska (٢٠٢٤) highlights that arbitral awards are rarely published, and if they are, they appear in a heavily redacted form that initially harms machine learning capabilities and, more than that, structurally constrains the data, giving disproportionate advantages to the actors holding proprietary internal archives, which can be considered a concentration of knowledge and power.

This means that the actors best positioned to benefit from AI are not necessarily those with the strongest legal arguments, but those with the greatest access to proprietary databases, internal archives, and computational resources. In practice, this tends to favor major arbitral institutions, large international law firms, repeat players, and technology providers predominantly located in the Global North. The resulting asymmetry is not merely informational; it is epistemic. It affects who can predict outcomes, identify patterns, model arbitrator behavior, and optimize procedural strategy.

In this respect, the integration of AI into arbitration resembles what Couldry and Mejias (2018) describe as data colonialism: a process through which social life is appropriated through data extraction and converted into asymmetrical forms of value. Mann and Daly (2019) likewise show that data-driven infrastructures often reproduce North–South relations of domination. Applied to arbitration, these insights suggest that AI may intensify the preexisting concentration of legal knowledge and strategic capacity within the arbitral marketplace.

A second asymmetry concerns vendor control and the increasing dependence of arbitral practice on privately owned technological infrastructures. As AI-assisted arbitration develops, the relevant tools are not being built primarily by arbitrators or public institutions, but by commercial actors operating through proprietary platforms, subscription databases, cloud environments, and integrated legal-information ecosystems. This matters because the exercise of procedural and strategic power may gradually shift toward those who control the architecture through which arbitral work is performed. In practical terms, the market is shaped not only by general-purpose technology providers, including large Global North digital firms active in cloud computing, enterprise software, and AI development, but also by specialized legal-information and arbitration-focused actors such as Thomson Reuters, Wolters Kluwer, and Jus Mundi. By way of illustration, Thomson Reuters presents artificial intelligence as part of its legal-information ecosystem; Wolters Kluwer similarly frames AI as central to its professional services strategy; and Jus Mundi markets Jus AI specifically within the international law and arbitration space.

These companies do not formally decide disputes, but they increasingly mediate access to authorities, analytics, workflow tools, research environments, and structured arbitral knowledge. In that sense, their role is infrastructural: they shape the informational conditions under which parties, counsel, institutions, and tribunals prepare, organize, and evaluate claims.

One way to reduce vendor control would be for countries in the Global South to establish regulatory sandboxes (Gromova & Ferreira, 2025). These sandboxes could be within public institutions, private arbitration providers, or associations to develop AI technologies for arbitration, which would be limited to that jurisdiction or provider and protect digital sovereignty principles. A regulatory sandbox is a secure testing environment with relaxed regulations, first introduced in 2014 when the UK Financial Conduct Authority (FCA) launched the first FinTech sandbox. Since then, many regulatory sandboxes have been established in both the Global North and the Global South. Therefore, creating new AI technologies for arbitration and other tools within a specialized regulatory environment makes sense, as arbitration is part of a country’s domestic jurisdiction embedded in the concept of a multi-door courthouse system (Sanders, 1976).

It is also important to distinguish between general-purpose AI infrastructure, often developed by large technology companies, and arbitration-focused AI applications built into legal information and dispute-resolution platforms, since arbitral dependency may arise from interactions between these layers.

This development is significant because these providers increasingly operate not as isolated software vendors but as integrated legal-information ecosystems. Their services may combine databases, analytics, drafting support, workflow management, search functions, and AI-assisted knowledge retrieval within a single proprietary environment. As a result, arbitral practice may become dependent on vertically integrated infrastructures that shape not only access to information but also the filtering, prioritization, and operationalization of legal knowledge. In practical terms, this means procedural sophistication may increasingly be tied to subscription access, platform familiarity, and purchasing power, thereby reinforcing inequalities between repeat players and less-resourced parties, including many actors from the Global South.

The literature on algorithmic governance helps explain why this form of private technological power should be taken seriously. Pasquale (2017) warns that opaque digital systems may undermine attribution, responsibility, and explainability precisely because they influence outcomes without being fully visible or contestable. Fortes (2020) similarly argues that digital decision-support systems must be critically assessed in light of transparency, legitimacy, and due process, rather than being treated as inherently reliable simply because they are technologically sophisticated. In the arbitral context, the concern is not limited to hypothetical robot arbitrators. As Kasap (2021) shows, the deeper problem lies in how AI-assisted systems may affect fact organization, legal reasoning, and the conditions of independent adjudicative judgment. The same concern appears in the arbitration-specific literature on large language models, where Tsuvina and Tsuvina (2024) note that AI tools may appear persuasive while remaining probabilistic, opaque, and difficult to verify. Once such systems are embedded in research, drafting, document review, and strategic forecasting, vendor-controlled infrastructures can influence arbitral practice long before any formal decision is rendered.

This creates a risk of technological dependency, especially from a Global South perspective. The more arbitration relies on subscription-based legal platforms, proprietary analytics, and privately curated AI environments, the more access to procedural sophistication becomes tied to market position and purchasing power rather than to legal merit alone. This may produce a form of infrastructural lock-in, in which the actors with the greatest financial and technological resources gain disproportionate advantages in speed, organization, predictive capacity, and informational reach. The problem is therefore not simply market concentration in legal tech; it is the fusion of arbitral governance with privately controlled systems that are disproportionately developed, financed, and governed in the Global North. Read through a TWAIL III lens, vendor control becomes part of a broader pattern in which technological infrastructure, like arbitral rulemaking and epistemic authority, is externalized from the Global South and reimported on asymmetrical terms. In that setting, private technological power does not replace older hierarchies in international arbitration; it digitizes and deepens them.

A third asymmetry lies in the production of normative standards governing AI in arbitration. The current soft law and institutional guidance on the use of AI in arbitration have emerged predominantly from organizations, institutions, and epistemic communities located in the Global North. This is evident in initiatives such as the CIArb Guideline on the Use of AI in Arbitration (Chartered Institute of Arbitrators, 2025), the SVAMC Guidelines on the Use of Artificial Intelligence in Arbitration (Silicon Valley Arbitration & Mediation Center, 2024), and institutionally framed experiments such as the AAA AI Arbitrator (American Arbitration Association, n.d.). These instruments are important and often sophisticated, but they also reveal who currently has the authority to define responsible AI use in global dispute resolution.

Bianchi (2020) shows that international arbitration is shaped by epistemic communities whose influence extends well beyond formal sources of law. Through best practices, professional discourse, institutional prestige, and repeated interaction among elite actors, these communities shape how problems are defined and which solutions appear legitimate. From a TWAIL-informed perspective, the key concern is not simply that these standards originate in the Global North, but that they may universalize assumptions developed in Global North professional and technological environments as if they were globally neutral. This reproduces the broader concern raised by Chimni (2006) and Pahuja (2011): transnational legal governance often presents itself as technical and universal while remaining deeply embedded in asymmetrical global structures. Although China offers an important partial counterpoint through initiatives such as CIETAC’s engagement with AI governance, the dominant regulatory frameworks still emerge from Northern arbitral and technological ecosystems.

While the SVAMC Guidelines on the Use of Artificial Intelligence in Arbitration and the CIArb Guideline on the Use of AI in Arbitration make an important contribution by addressing transparency, human responsibility, confidentiality, and procedural fairness, they also illustrate how the governance of AI in arbitration is being shaped predominantly by institutions and epistemic communities located in the Global North (Chartered Institute of Arbitrators, 2025; Silicon Valley Arbitration & Mediation Center, 2024). Both instruments are framed as neutral and practical responses to technological change, emphasizing that AI should be used in a manner consistent with due process, party autonomy, and the integrity of arbitral proceedings.

Nevertheless, their very existence as leading reference points in the field reveals a deeper asymmetry: the normative frameworks that define responsible AI in arbitration are not being produced through genuinely plural or globally representative processes, but through professional networks centered in London, Silicon Valley, and adjacent legal-technological hubs. Their authority, therefore, derives not only from the quality of their recommendations but also from the prestige and institutional centrality of the communities producing them, which raises questions of representational legitimacy in a field that presents itself as global.

In this sense, the issue is not that these guidelines are normatively misguided; many of their recommendations are sensible and necessary. The problem is that they consolidate the Global North’s position not only as the primary producer of arbitral technologies, but also as the principal author of the soft law that governs those technologies. As Bianchi (2020) suggests, epistemic communities exercise influence by determining which ideas become authoritative in international arbitration, and these guidelines are a contemporary example of that process in the AI context.

The importance of BRICS initiatives in this context lies not just in diversifying institutions but also in their potential to shift norm-making, technological testing, and arbitral governance beyond the traditional Global North hub of international dispute resolution. Early proposals, like the BRICS Supranational Arbitration Center (2014), gradually developed into more decentralized and collaborative models, including the BRICS Arbitration Institute proposal (2021), the BRICS Arbitration Court launched by the Guangzhou Arbitration Commission in 2023, and the BRICS+ Arbitration Mechanism (BRICSAM) suggested by the Russian Arbitration Center in 2024. Instead of creating a single centralized body, these efforts increasingly favor a network-based structure where national arbitration institutions manage proceedings together under shared rules and coordinated governance frameworks. Although these efforts still face significant issues such as legal fragmentation, enforcement uncertainties, institutional disparities, geopolitical tensions, and economic sanctions, they show a strategic move toward stronger cooperation among BRICS countries.

In this way, the growth of BRICS and the enhancement of arbitration collaboration could be an important step in reducing the long-standing dominance of Global North institutions in international dispute resolution (Fatima, 2022). Even though these efforts are still in early stages, they are meaningful because they indicate that institutional cooperation outside traditional Northern arbitral centers can promote not only greater diversity in forums but also a wider range of approaches to technological governance in the changing world of global dispute resolution.

From a TWAIL-oriented perspective, this dynamic is especially significant because it reflects what this article describes as an emerging TWAIL III moment: the extension of North–South hierarchy into digital infrastructures, technical standards, and algorithmic governance. The SVAMC and CIArb texts seek to discipline AI use. However, they do so from within institutional environments that already possess disproportionate access to data, vendors, technical expertise, and reputational capital. As a result, the governance of AI risks reproducing what Couldry and Mejias (2018) call data colonialism.

The Global South is thus placed in a structurally subordinate position: not only is it less likely to control the relevant datasets and technological platforms, but it is also less likely to shape the norms by which those platforms are assessed and legitimized. Even where the guidelines invoke universal values such as fairness, transparency, and accountability, they may still function as vehicles of normative dependency by universalizing regulatory assumptions developed in Global North settings. Read in this way, the guidelines are not merely instruments of best practice; they are also part of a broader political economy of arbitral digitalization in which the Global North retains leadership over infrastructure, knowledge production, and rule-making, thereby reinforcing new forms of digital colonialism within transnational dispute resolution.

This broader trend is also evident in transnational commercial harmonization. UNCITRAL has already begun addressing automation and artificial intelligence through the UNCITRAL Model Law on Automated Contracting (2025), which expressly recognizes the “increasing relevance of automation in contracting, including through the deployment of artificial intelligence systems” and defines, in Article 1 (1a), an “automated system” as a computer system capable of carrying out actions without the necessary review or intervention of a natural person (UNCITRAL, 2025). The Model Law further clarifies, in Article 1 (2), that automated systems may operate in deterministic or non-deterministic ways, addresses attribution and unexpected actions, and preserves the role of information requirements regarding the design, operation, or use of automated systems (UNCITRAL, 2025). Although this instrument is not arbitration-specific, it is highly relevant because it shows that automated systems and AI are no longer peripheral to transnational commercial law. They are already being treated as a subject of harmonized legal regulation. This demonstrates that AI-related automation is moving from the margins of legal experimentation into the mainstream vocabulary of transnational commercial regulation.

For this article, the UNCITRAL initiative is significant in two ways. First, it confirms that AI-related automation is becoming normalized within the legal architecture of cross-border commerce, which will inevitably affect dispute resolution environments, including arbitration. Second, it reveals that the governance of automation is increasingly framed through technology-neutral and market-facilitating instruments that aim to provide legal certainty for digital trade. That development is valuable, but it also underscores the need for a more explicit distributive analysis. The legal recognition of automated systems may promote efficiency and predictability. However, it does not by itself resolve who controls those systems, who benefits from their deployment, or whether parties from the Global South will participate in this new technological order on equal terms. In that sense, even global harmonization efforts must be read alongside the broader TWAIL III concern that emerging legal frameworks for AI and automation may universalize technologically advanced models of commerce without adequately confronting their asymmetrical material foundations.

Accordingly, AI governance in arbitration must be understood as part of a broader structure of private transnational ordering. It is not limited to the use of smart tools by individual lawyers or tribunals. Rather, it concerns who controls the data, who owns the infrastructure, who drafts the standards, and whose knowledge becomes authoritative in shaping the future of dispute resolution. In that sense, AI does not enter arbitration as a neutral innovation. It enters an already concentrated field and may strengthen the existing nexus between arbitral governance, technological dependency, and epistemic hierarchy.

3.2. Procedural Integrity and Safeguards in AI-Assisted Arbitration

These structural concerns are important because they have direct procedural implications. If some parties have access to better predictive tools, larger databases, and more advanced AI-assisted document and argument analysis, equality of arms may be compromised even if formal procedural rights stay the same. Similarly, if arbitrators, counsel, or institutions rely on AI tools without enough disclosure, parties might be affected by technological mediation they cannot recognize, understand, or challenge. Fortes (2020), writing on digital justice and due process, argues that automated and algorithmic systems need procedural guarantees that let affected parties examine the logic and impact of digital decision-support mechanisms. Although arbitration is not the same as public courts, the core concern is similar: procedural fairness is threatened when an influential technological layer remains hidden, unchallengeable, or immune from genuine scrutiny.

This matters especially because AI can influence arbitral procedures at many levels without ever formally deciding the case. It can affect document production, legal research, internal drafting, witness preparation, settlement evaluation, and strategic forecasting. In such cases, the line between support and influence becomes more unclear. Gicquello (2020) notes that AI might, in some ways, improve arbitral legitimacy, fairness, and efficiency, but that depends on how it is introduced and managed. The same technology that can reduce human error or improve consistency could also increase bias, embed unequal assumptions, or hide relevant reasoning behind non-transparent computational processes. Kasap (2021) emphasizes that using AI in adjudication raises concerns not just about substantive bias but also about fact-finding, reasoning, and the overall integrity of decision-making. Therefore, the issue is not whether AI is inherently good or bad, but whether its use is protected by safeguards that maintain party equality, due process, and adjudicative accountability.

At a minimum, this requires a procedural integrity framework adapted to AI-assisted arbitration. Such a framework should include, first, transparency obligations regarding whether AI tools are used, who uses them, and for what purpose. Second, it should incorporate a proportional principle of explainability: the greater the AI’s influence on strategy, evidence organization, or adjudicative reasoning, the stronger the justification for disclosure and scrutiny. Third, parties should have some degree of contestability, meaning a meaningful opportunity to challenge the use of AI where it might affect fairness, confidentiality, evidentiary balance, or the integrity of deliberation. This concern is especially important with large language models, whose outputs may seem persuasive but remain probabilistic, non-deterministic, and susceptible to hallucination or hidden bias (Tsuvina & Tsuvina, 2024). Fourth, confidentiality and data protection must stay central, particularly when parties upload sensitive documents to third-party systems or when data could be retained, reused, or processed beyond the parties’ expectations. Fifth, human oversight must remain essential. Even if AI tools support drafting, information management, or predictive assessment, responsibility for procedural decisions and final adjudicative results must remain with human decision-makers.

In practice, these safeguards could be implemented through procedural orders, institutional rules, or soft-law guidance requiring disclosure of material AI use by parties, counsel, arbitrators, or institutions; identification of the functional role played by the system; preservation of confidentiality and data provenance; and providing an opportunity for the opposing party to object if AI use might compromise procedural fairness. While these measures would not eliminate all asymmetries, they would help reduce the risk that AI functions as an invisible layer of influence in arbitral proceedings.

In this regard, the most convincing approach is not to reject AI in arbitration altogether, but to reject unregulated AI. AI use should still align with fundamental procedural guarantees such as the right to be heard, equal treatment, impartial decision-making, and the obligation to provide reasons. Fortes (2020) and Pasquale (2017) are especially helpful here because both emphasize, in different ways, that due process in algorithmically mediated settings requires transparency, accountability, and opportunities for challenge. Their insights are highly relevant in the arbitral context. If a party cannot determine that AI was used, cannot understand its role, and cannot object if that role impacts procedural fairness, then the value of arbitral due process is greatly diminished. Similarly, legal technologies affect not only the speed of dispute resolution but also the framework within which authority is exercised. Therefore, AI cannot be regarded as a neutral procedural tool; it must be considered an integral part of the adjudication structure.

Existing soft-law initiatives already point in this direction, even if they do not fully resolve the structural inequalities identified in this article. In this way, the CIArb Guideline on the Use of AI in Arbitration (Chartered Institute of Arbitrators, 2025) is an important effort to integrate AI into arbitration while ensuring the parties’ autonomy, fairness, and ethical standards. It presents the benefits and risks of using AI in arbitration, offers general recommendations, and includes, in Appendix A, a template agreement on AI use in arbitration and, in Appendix B, a template procedural order for its application. The institution highlights risks, including threats to confidentiality, data integrity, impartiality, independence, and the enforceability of arbitral awards. This underscores the importance of disclosing the use of an AI tool to maintain the integrity of the proceedings.

The arbitrator must, when necessary, impose AI-related disclosure requirements on the parties through a case-by-case analysis as outlined in Guideline 3 of the Guidelines on the Use of Artificial Intelligence in Arbitration (Silicon Valley Arbitration & Mediation Center, 2024). Łągiewska (٢٠٢5) further develops this disclosure logic, proposing a tiered, color-coded framework modeled on the IBA Guidelines on Conflicts of Interest that distinguishes among prohibited uses (red list, non-waivable), uses permissible only with the parties’ consent (red list, waivable), uses that merely warrant disclosure (orange list), and uses for which there is no obligation to disclose (green list). She illustrates the stakes through the LaPaglia dispute, in which an arbitrator’s undisclosed reliance on AI was challenged. Conversely, all decisions made by arbitrators must be human, meaning AI tools should serve as supporting aids rather than replacing legal analysis, interpretation, or the application of facts and law. An arbitrator is responsible for the decision rendered; therefore, he cannot delegate any part of the decision-making process to AI.

Major Brazilian arbitral institutions have so far adopted a cautious and transparency-focused approach to artificial intelligence. While encouraging participants in arbitral proceedings to discuss and disclose the use of AI tools when appropriate, institutions like CAM-CCBC and CAM/Amcham have explicitly stated in recent administrative resolutions that they do not currently use artificial intelligence in managing proceedings, due to the range of risks associated with such technologies, including concerns about confidentiality, procedural fairness, and governance (CAM-CCBC, 2025; CAM/Amcham, 2025). This development is notable, as it indicates that leading Brazilian providers acknowledge both the practical importance of AI and the need for institutional caution until stronger safeguards are established.

From the perspective of the Global South, these safeguards must go beyond generic compliance language. They should be crafted to address structural inequality in access to technology, datasets, and influence over standard-setting. Otherwise, AI governance in arbitration risks becoming another example of what this article has highlighted as an emerging third TWAIL moment: a form of transnational ordering where historical North–South hierarchies are redefined through digital infrastructures, computational power, and epistemic control. In this context, the issue is not just that the Global South may have limited access to AI tools; it is that these tools may be integrated into an arbitral environment in which the relevant tools, datasets, standards, and governance models are predominantly created elsewhere. Therefore, the future of AI in arbitration should not only focus on innovation and efficiency but also emphasize institutional justice, technological sovereignty, and the equitable distribution of epistemic power in global dispute resolution.

4. Conclusion

This article argues that the increasing integration of artificial intelligence into international arbitration should not be viewed solely as a technological advancement aimed at improving efficiency, reducing costs, or expediting proceedings. Instead, AI must be understood within the broader political economy of international dispute resolution, a field already influenced by market concentration, private governance, and epistemic hierarchy. From this perspective, the key question is whether the spread of AI in arbitration risks reinforcing and extending the Global North’s dominance over the Global South. The analysis in this article supports a nuanced affirmative answer. This does not mean that AI is inherently incompatible with arbitration, nor that all initiatives led by the Global North are necessarily problematic. Rather, it highlights that the distributive and structural conditions under which AI is created, deployed, and governed must be acknowledged if technological progress is not to perpetuate existing hierarchies. AI undoubtedly offers significant practical benefits, including increased speed, consistency, organization, and access to information. However, these benefits are being introduced into an arbitral system that is already structurally dominated by a limited number of institutions, seats, professional networks, and technological ecosystems historically tied to the Global North.

The approach demonstrates that this risk manifests through at least three interconnected mechanisms. First, AI relies on data, which remains unevenly distributed, often concentrated among major institutions, large law firms, repeat players, and platform providers. Second, the growing importance of vendor control indicates that arbitral practices are being mediated by privately owned technological infrastructures, including legal information ecosystems and AI-powered platforms developed primarily by companies based in the Global North. Third, the standards for responsible AI use in arbitration are also mainly emerging from institutions and epistemic communities in the Global North, meaning that the rules governing technological change are created asymmetrically. Collectively, these developments suggest that AI could extend not only arbitration’s efficiency but also the epistemic, infrastructural, and normative advantages already held by actors at the core of the global arbitral system.

Therefore, the article argues that the relationship between AI, arbitration, and the Global South is best understood through the framework of an emerging TWAIL III. TWAIL I highlighted how the Third World was excluded from shaping classical international law, and TWAIL II exposed how dominance became embedded in modern international legal institutions and doctrines. The current moment marks a further stage in which hierarchy is reproduced through digital infrastructures, proprietary systems, data concentration, and algorithmic governance. In the arbitral context, this means that governing AI is not just about managing innovation; it also involves who controls the tools, defines the standards, and whose knowledge is deemed authoritative. The rise of AI-supported arbitration, therefore, risks becoming another arena where the Global South participates as a user of systems largely designed, owned, and regulated elsewhere.

While this article does not argue that AI should be rejected in arbitration, it emphasizes that AI in arbitration must be governed in ways that are procedurally fair, institutionally transparent, and distributively conscious. The challenge lies not in technological modernization itself but in the uncritical adoption of technology within a structurally unequal transnational legal system. Therefore, procedural safeguards such as disclosure of AI use, explainability proportional to the procedural impact, contestability, data protection, confidentiality, and strong human oversight are essential. However, safeguards alone are not enough if they are only framed in generic or formal terms. A truly fair AI governance model for arbitration must also address the unequal distribution of technological infrastructure, market power, and epistemic authority between the Global North and the Global South.

Ultimately, the future of AI in international arbitration should not be judged solely based on efficiency, speed, or user satisfaction. Instead, it should be evaluated on whether it promotes a more equitable and diverse system of global dispute resolution or perpetuates existing inequalities under the guise of innovation. If AI governance continues to be shaped mainly by the same institutions, vendors, and professional communities that already dominate international arbitration, then the digital shift may reinforce rather than challenge arbitral hierarchies. Conversely, if AI governance is approached critically and inclusively, with attention to technological sovereignty, representational diversity, and procedural integrity, AI may be integrated into arbitration without reproducing the structural patterns of historical North–South subordination identified in this article. Unless those structural and distributive conditions are made visible and addressed, the digital transformation of arbitration is more likely to consolidate existing hierarchies than to democratize global dispute resolution.

This study also notes certain limitations. The analysis is mainly qualitative and conceptual, relying on doctrinal literature, institutional reports, and policy documents rather than large-scale empirical data on the use of artificial intelligence in arbitration. Because arbitration proceedings are confidential and the use of AI tools is still developing, systematic empirical evidence remains limited. Additionally, the article primarily focuses on major arbitral institutions and global technology providers, which might not fully reflect practices in smaller or emerging arbitration centers, especially in the Global South. Future research could supplement this approach with empirical and interdisciplinary studies.

Future research should therefore explore institutional mechanisms that allow Global South arbitral institutions, regulators, and scholars to take a more active role in shaping AI governance in international dispute resolution.

Bibliographic references

American Arbitration Association. (n.d.). AI arbitrator. https://www.adr.org/ai-arbitrator/

Anand, R. P. (1962). Rôle of the “New” Asian-African Countries in the Present International Legal Order. American Journal of International Law, 56(2), 383–406. https://doi.org/10.2307/2196445

Anand, R. P. (1966a). Attitude of the Asian-African States Toward Certain Problems of International Law. International and Comparative Law Quarterly, 15(1), 55–75. https://doi.org/10.1093/iclqaj/15.1.55

Anand, R. P. (1966b). Sovereign Equality of States in International Law—II. International Studies, 8(4), 386-421.

Anand, R. P. (1967). Sovereign equality of states in international law—I. International Studies, 8(3), 213–233.

Anghie, A. (1996). Francisco de Vitoria and the colonial origins of international law. Social & Legal Studies, 5(3), 321–336.

Berger, K. P. (1993). International economic arbitration. Kluwer Law International.

Bianchi, A. (2020). Epistemic communities in international arbitration. In J. d’Aspremont & S. Singh (Eds.), Concepts for international law: Contributions to disciplinary thought (pp. 569–590). Oxford University Press. https://doi.org/10.1093/law/9780198796190.003.0024

Centro de Arbitragem e Mediação da Amcham Brasil. (2025). Resolução administrativa n.º 05/2025. https://mkt.amcham.com.br/materiais/cam/resolucao-adm-05-2025.pdf

Centro de Arbitragem e Mediação da Câmara de Comércio Brasil-Canadá. (2025). Orientação administrativa n.º 07/2025. https://www.ccbc.org.br/cam-ccbc-centro-arbitragem-mediacao/orientacao-administrativa-no-07-2025/

Chartered Institute of Arbitrators. (2025). Guideline on the use of AI in arbitration (Updated September 2025). https://www.ciarb.org/media/bpndtcgu/guideline-on-the-use-of-ai-in-arbitration_updated-sept-2025.pdf

Chimni, B. S. (2004). International institutions today: An imperial global state in the making. European Journal of International Law, 15(1), 1–37. https://doi.org/10.1093/ejil/15.1.1

Chimni, B. S. (2006). Third World approaches to international law: A manifesto. International Community Law Review, 8(1), 3–27. https://doi.org/10.1163/187197306779173220

China International Economic and Trade Arbitration Commission. (2025). CIETAC explores the use of artificial intelligence in arbitration. https://www.cietac.org/articles/33752

Couldry, N., & Mejias, U. A. (2018). Data colonialism: Rethinking big data’s relation to the contemporary subject. Television & New Media, 20(4), 336–349. https://doi.org/10.1177/1527476418796632

Cox, R. W. (1981). Social forces, states and world orders: Beyond international relations theory. Millennium: Journal of International Studies, 10(2), 126–155. https://doi.org/10.1177/03058298810100020501

Drahozal, C. R. (2000). Commercial norms, commercial codes, and international commercial arbitration. Vanderbilt Journal of Transnational Law, 33, 79–113.

Drahozal, C. R. (2004). Regulatory competition and the location of international arbitration proceedings. International Review of Law and Economics, 24(3), 371–384. https://doi.org/10.1016/j.irle.2004.10.002

Fatima, N. (2022). Future of dispute resolution and investment in BRICS. Revista Brasileira de Alternative Dispute Resolution (RBADR), 4(8), 187–209. https://doi.org/10.52028/rbadr.v4i8.9

Ferreira, D. B., & Gromova, E. (2025). The bossware era and the e-panopticon: Current technologies and legal challenges. Industrial Law Journal, 55(1), 181-212. https://doi.org/10.1093/indlaw/dwaf006

Fortes, P. R. B. (2020). Paths to digital justice: Judicial robots, algorithmic decision-making, and due process. Asian Journal of Law and Society, 7(3), 453–469. https://doi.org/10.1017/als.2020.12

Foucault, M. (2013). The archaeology of knowledge. Routledge.

Galindo, G. R. B. (2017). Splitting TWAIL? Windsor Yearbook of Access to Justice, 33(3), 37–56. https://doi.org/10.22329/wyaj.v33i3.4886

Gathii, J. T. (2011). TWAIL: A brief history of its origins, its decentralized network, and a tentative bibliography. Trade, Law and Development, 3(1), 26–64.

Gicquello, M. (2020). Artificial intelligence in international arbitration. In T. Schultz & F. Ortino (Eds.), Concepts for international law: Contributions to disciplinary thought (pp. 591–616). Oxford University Press. https://doi.org/10.1093/law/9780198796190.003.0025

Grimm, D. (2015). Sovereignty: The origin and future of a political and legal concept. Columbia University Press.

Gromova, E., & Ferreira, D. B. (2024). On the way to BRICS+ digital sovereignty: Opportunities and challenges of a new era. BRICS Law Journal, 11(3), 54–69. https://doi.org/10.21684/2412-2343-2024-11-3-54-69

Gromova, E. A., & Ferreira, D. B. (2025). Regulatory sandbox: Bridging the gaps by designing a dispute resolution system. Conflict Resolution Quarterly, 43(1), 91–98. https://doi.org/10.1002/crq.21482

Horton, D. (2020). The arbitration rules: Procedural rulemaking by arbitration providers. Minnesota Law Review, 105, 619–678.

International Centre for Settlement of Investment Disputes. (2024). The ICSID caseload: Statistics. World Bank Group. https://icsid.worldbank.org/resources/publications/icsid-caseload-statistics

International Chamber of Commerce. (2024). ICC dispute resolution statistics 2024. https://iccwbo.org/news-publications/news/icc-dispute-resolution-statistics-2024/

International Chamber of Commerce. (n.d.). ICC International Court of Arbitration: Court members. Retrieved March 15, 2026, from https://iccwbo.org/dispute-resolution/dispute-resolution-services/icc-international-court-of-arbitration/court-members/

Jus Mundi. (n.d.). Jus AI. Retrieved March 14, 2026, from https://jusmundi.com/en/jus-ai

Karlstad, W. (2023, May 17). Explore the path to digital sovereignty! Tietoevry. https://www.tietoevry.com/en/blog/2023/05/all-you-need-to-know-about-digital-sovereignty/

Kasap, G. H. (2021). Can artificial intelligence (“AI”) replace human arbitrators? Technological concerns and legal implications. Journal of Dispute Resolution, 2021(2), 209-254. https://doi.org/10.31235/osf.io/k4g8s

Łągiewska, M. (٢٠٢٤). New technologies in international arbitration: A game-changer in dispute resolution? International Journal for the Semiotics of Law, 37(3), 851–864. https://doi.org/10.1007/s11196-023-10070-7

Łągiewska, M. (٢٠٢5). Artificial intelligence and international arbitration law: Revolution or evolution. Routledge. https://doi.org/10.4324/9781003667834

Landes, W. M., & Posner, R. A. (1979). Adjudication as a private good. The Journal of Legal Studies, 8(2), 235–284. https://doi.org/10.1086/467609

London Court of International Arbitration. (2025). LCIA annual casework report 2024. Retrieved March 14, 2026, from https://www.lcia.org/News/lcias-2024-annual-casework-report.aspx

Mann, M., & Daly, A. (2019). (Big) data and the North-in-South: Australia’s informational imperialism and digital colonialism. Television & New Media, 20(4), 379–395. https://doi.org/10.1177/1527476418806091

McLaughlin, M. R. (2023). The geoeconomics of Belt and Road disputes: A case study on the China–Pakistan Economic Corridor. Asian Journal of International Law, 14(1), 94–122. https://doi.org/10.1017/S2044251323000176

Mutua, M. (2000). What is TWAIL? Proceedings of the ASIL Annual Meeting, 94, 31–38. https://doi.org/10.1017/S0272503700054896

Najar, J. C. (2009). Inside out: A user’s perspective on challenges in international arbitration. Arbitration International, 25(4), 515–528.

Okafor, O. C. (2008). Critical Third World approaches to international law (TWAIL): Theory, methodology, or both? International Community Law Review, 10, 371–378.

Pahuja, S. (2011). Decolonising international law: Development, economic growth and the politics of universality. Cambridge University Press.

Pasquale, F. (2017). Toward a fourth law of robotics: Preserving attribution, responsibility, and explainability in an algorithmic society. Ohio State Law Journal, 78(5), 1243–1255.

Rajagopal, B. (2003). International law from below: Development, social movements, and Third World resistance. Cambridge University Press.

Santos, B. de S. (2002). Toward a sociology of absences and a sociology of emergences. Revista Crítica de Ciências Sociais, 63, 237–280.

Santos, B. de S. (2007). Beyond abyssal thinking: From global lines to ecologies of knowledges. Review (Fernand Braudel Center), 30(1), 45–89.

Schneider, N. (2022). Governable stacks against digital colonialism. tripleC: Communication, Capitalism & Critique, 20(1), 19–36. https://doi.org/10.31269/triplec.v20i1.1281

Silicon Valley Arbitration & Mediation Center. (2024). SVAMC guidelines on the use of artificial intelligence in arbitration (1st ed.). https://svamc.org/wp-content/uploads/SVAMC-AI-Guidelines-First-Edition.pdf

Singapore International Arbitration Centre. (2024). SIAC annual reports. https://siac.org.sg/annual-reports

Solhchi, M. A., & Baghbanno, F. (2023). Artificial intelligence and its role in shaping the future of arbitration. International Journal of Law in Changing World, 2(2), 56–76. https://doi.org/10.54934/ijlcw.v2i2.56

Susskind, J. (2018). Future politics: Living together in a world transformed by tech. Oxford University Press.

The GeoEconomics. (n.d.). Singapore’s dual identity: Heart of Global South and soul of North. Retrieved March 13, 2026, from https://thegeoeconomics.com/singapores-dual-identity-heart-of-global-south-and-soul-of-north/

Thomson Reuters. (n.d.). Introduction to artificial intelligence at Thomson Reuters. https://www.thomsonreuters.com/en/artificial-intelligence/introduction-to-artificial-intelligence-at-thomson-reuters

Tsuvina, T., & Tsuvina, A. (2024). Rethinking regulation: Integrating large language models in international arbitration. Problems of Legality, (165), 212–225. https://doi.org/10.21564/2414-990x.166.315451

UNCITRAL. (2025). UNCITRAL Model Law on Automated Contracting. United Nations. https://uncitral.un.org/en/mlac

United Nations Commission on International Trade Law. (n.d.). Status: Convention on the Recognition and Enforcement of Foreign Arbitral Awards (New York, 1958). Retrieved March 5, 2026, from https://uncitral.un.org/en/texts/arbitration/conventions/foreign_arbitral_awards/status2

United Nations Commission on International Trade Law. (n.d.). Status: UNCITRAL Model Law on International Commercial Arbitration. Retrieved March 7, 2026, from https://uncitral.un.org/en/texts/arbitration/modellaw/commercial_arbitration/status

White & Case LLP, & School of International Arbitration, Queen Mary University of London. (2015). 2015 international arbitration survey: Improvements and innovations in international arbitration. https://www.qmul.ac.uk/arbitration/media/arbitration/docs/2015_International_Arbitration_Survey.pdf

White & Case LLP, & School of International Arbitration, Queen Mary University of London. (2018). 2018 international arbitration survey: The evolution of international arbitration. https://www.qmul.ac.uk/arbitration/media/arbitration/docs/2018-International-Arbitration-Survey---The-Evolution-of-International-Arbitration-(2).PDF

White & Case LLP, & School of International Arbitration, Queen Mary University of London. (2021). 2021 international arbitration survey: Adapting arbitration to a changing world. https://www.qmul.ac.uk/arbitration/media/arbitration/docs/LON0320037-QMUL-International-Arbitration-Survey-2021_19_WEB.pdf

White & Case LLP, & School of International Arbitration, Queen Mary University of London. (2025). 2025 international arbitration survey: The path forward—Realities and opportunities in arbitration. https://www.qmul.ac.uk/arbitration/media/arbitration/docs/White-Case-QMUL-2025-International-Arbitration-Survey-report.pdf

Whytock, C. A. (2008). Litigation, arbitration, and the transnational shadow of the law. Duke Journal of Comparative & International Law, 18, 449–475.

Wolters Kluwer. (n.d.). Artificial intelligence. Retrieved March 14, 2026, from https://www.wolterskluwer.com/en/about-us/artificial-intelligence

Yıldırım, A. C. (2024). The use of technology in case management in international investment arbitration: A realistic approach. Arbitration International, 40(2), 233–250. https://doi.org/10.1093/arbint/aiae010

(*) Nota del Equipo Editorial: Este artículo fue recibido el 15 de marzo de 2026 y su publicación fue aprobada el 14 de junio de 2026.

(**) AI use statement: Artificial intelligence tools were used only for limited language assistance in the preparation of this manuscript, including translation from Portuguese into English and editing to improve grammar, clarity, style, and proofreading. No AI tools were used to generate the article’s arguments, legal analysis, conclusions, or scholarly claims. The authors take full responsibility for the content of the manuscript.

(***) Lawyer from the Pontifical Catholic University of Rio de Janeiro (Rio de Janeiro, Brazil). Ph.D. in Legal Sciences, Constitutional Law, and State Theory from the same university. Editor-in-Chief of the “Brazilian Journal of Alternative Dispute Resolution” (RBADR) and Co-Editor-in-Chief of the “International Journal of Law in Changing World” (IJLCW). Fellow of the Chartered Institute of Arbitrators (CIArb). Founding partner of DBFLaw. ORCID: https://orcid.org/0000-0003-0504-1154. E-mail: daniel@dbflaw.com.br.

(****) Lawyer from the South Ural State University (Chelyabinsk, Russia). Doctor of Law from the same university. Professor at the Department of Civil Law, South Ural State University. ORCID: https://orcid.org/0000-0001-6655-8953. E-mail: gromovaea@susu.ru.