Tarana SHAHBAZOVA (Azerbaijan),
 international relations specialist 

Summary. This article examines the political and legal challenges of artificial intelligence regulation in the context of transforming international relations and the global governance system. It explores the impact of artificial intelligence on state sovereignty and strategic competition, the contradictions between national regulatory models, the challenges of establishing legal responsibility for AI-driven decisions, and the risks posed to human rights and democratic governance. Particular attention is paid to the fragmentation of international legal mechanisms and the need to strengthen multilateral cooperation. The article argues that effective AI regulation requires reconciling national interests with international obligations, ensuring transparency and accountability, and preserving meaningful human oversight of decisions affecting fundamental rights and freedoms. It concludes that a multilateral AI governance framework is necessary to reconcile technological development with the protection of fundamental rights and international responsibility.

Keywords: artificial intelligence, international relations, state sovereignty, global governance, international law, legal accountability, human rights, digital divide, international cooperation

Artificial intelligence as a new factor in global power politics: sovereignty, strategic competition, and the digital divide. Artificial intelligence is becoming a defining factor in contemporary international relations, transforming traditional understandings of power, sovereignty, and global competition. Whereas international influence was historically associated with military strength, economic resources, and territorial control, technological capabilities increasingly determine a state's ability to protect its interests, shape international rules, and maintain strategic autonomy. AI differs from earlier technological innovations because it does not merely improve existing instruments of power; it can also transform the processes through which political decisions are made, intelligence is analyzed, public opinion is influenced, and military operations are conducted. Consequently, competition over AI has become an integral component of the broader struggle for international influence.

One of the most important implications of this transformation concerns the relationship between technological dependence and national sovereignty. In the digital age, formal political independence does not necessarily guarantee technological autonomy. States may depend on foreign companies for advanced semiconductors, cloud infrastructure, foundational AI models, and essential software. Such dependencies can create vulnerabilities when access to critical technologies becomes subject to export restrictions, commercial decisions, or geopolitical pressure. The concept of AI sovereignty therefore extends beyond domestic ownership of technological infrastructure. It concerns a state's practical ability to make independent decisions about the development, deployment, and regulation of AI without becoming excessively vulnerable to external actors. For most countries, achieving this objective does not require complete technological self-sufficiency, but rather a combination of domestic expertise, diversified partnerships, resilient infrastructure, and effective regulatory institutions.

The technological rivalry between the United States and China illustrates the growing strategic significance of AI. Competition between these powers encompasses advanced computing, semiconductor production, scientific research, investment, and the establishment of technological standards. Their efforts to strengthen domestic capabilities demonstrate that AI is increasingly regarded as a strategic resource closely connected to economic security and international influence. Nevertheless, global leadership cannot be measured exclusively by the performance of individual AI models. It also depends on access to computing infrastructure, the quality of research institutions, the availability of skilled professionals, and the capacity to translate technological advances into economic and political advantages. Moreover, the concentration of these capabilities in major technology corporations introduces an additional dimension: private companies increasingly control resources essential to national security and public administration, creating complex relationships between commercial interests and state authority.

AI also changes the instruments through which political influence is exercised. Generative systems can produce persuasive content, imitate authentic communications, and facilitate the large-scale distribution of political messages. Although these technologies can improve public diplomacy and access to information, they may also support disinformation campaigns and foreign interference. In this environment, the ability to generate and distribute information becomes increasingly important to international competition. The principal concern is not simply the spread of fabricated content, but the potential erosion of public trust in the information institutions on which democratic decision-making and international diplomacy depend [1, p. 201].

At the same time, the unequal distribution of AI capabilities risks deepening existing global inequalities. The digital divide increasingly concerns not only access to the internet, but also access to advanced computing, research funding, technical expertise, and the ability to influence international standards. Countries that primarily consume technologies developed elsewhere may benefit from their applications while remaining dependent on external providers and having limited influence over the rules governing their use. These disparities also have a cultural and linguistic dimension: AI systems may inadequately represent societies whose languages, knowledge, and experiences are underrepresented in training data. As a result, technological inequality can become a source of economic disadvantage and reduced influence over the emerging digital order.

For countries such as Azerbaijan, these developments highlight the importance of combining technological modernization with strategic autonomy. Rather than attempting to reproduce every capability possessed by major AI powers, states can prioritize areas in which AI supports national development, strengthen domestic research and professional expertise, diversify technological partnerships, and participate actively in international standard-setting. Such an approach recognizes that sovereignty in the digital age depends not on isolation from global technological networks, but on the capacity to engage with them on informed and sustainable terms.

Ultimately, artificial intelligence is reshaping global power politics by connecting technological capability with economic competitiveness, political influence, and national sovereignty. Yet the concentration of AI resources and the unequal distribution of opportunities create a fundamental contradiction: technologies with global consequences remain substantially controlled by a limited number of states and private actors. The central geopolitical challenge is therefore not simply determining who will lead the AI revolution, but ensuring that technological leadership does not become an unrestricted instrument of political dominance or a permanent source of international inequality. This tension provides the foundation for examining the fragmented regulatory approaches and international legal coordination problems that characterize contemporary AI governance.

The fragmentation of ai governance: competing regulatory models and the crisis of international legal coordination. The rapid development of artificial intelligence has exposed a fundamental contradiction in contemporary global governance: although AI technologies operate across national borders, the rules governing their development and deployment remain largely shaped by divergent political priorities, legal traditions, and strategic interests. Algorithms developed in one jurisdiction can be integrated into services used worldwide, while the consequences of their operation may affect individuals, businesses, and public institutions located thousands of kilometres from their developers. Yet the international community has not established a comprehensive and universally binding legal framework capable of addressing the full range of risks associated with AI. The resulting fragmentation is not merely a technical obstacle to regulatory harmonization. It reflects deeper disagreements about the relationship between technological innovation, state authority, economic competition, and the protection of fundamental rights.

The European Union, the United States, and China illustrate three influential, although internally evolving, approaches to AI governance. The European Union has sought to establish a comprehensive, risk-based regulatory framework through the EU Artificial Intelligence Act, which entered into force in August 2024. Its central logic is that the permissibility of AI applications should depend partly on the risks they pose to health, safety, and fundamental rights. By imposing differentiated obligations and prohibitions, the EU seeks to establish safeguards while maintaining conditions for technological innovation. Its regulatory influence extends beyond European territory because international companies may adapt their products and compliance practices to retain access to the European market. This phenomenon, often described as the Brussels effect, demonstrates how regulatory authority can become a source of geopolitical influence even when it is not accompanied by technological dominance.

The United States has traditionally relied more heavily on private-sector innovation, market competition, sector-specific regulation, and executive policy initiatives. Although American institutions have developed various approaches to AI safety, civil rights, national security, and consumer protection, the overall framework has been less centralized than the EU model and has evolved with changes in political leadership. This approach reflects the view that excessive regulatory constraints may weaken innovation and undermine the country's competitive position, particularly in relation to China. However, a more flexible regulatory environment also raises questions about the consistency of safeguards, the accountability of technology companies, and the ability of existing legal institutions to address harms that do not fit neatly within established regulatory categories. The underlying tension is therefore between preserving the dynamism of technological development and ensuring that commercial incentives do not outweigh broader public interests [2, p.108].

China, in turn, has developed a regulatory approach that combines technological advancement, state supervision, cybersecurity, data governance, and control over certain forms of algorithmic activity. Its framework reflects the importance assigned to political stability, information management, and the state's authority over the digital environment. Rather than treating AI regulation primarily as a question of balancing individual rights against technological risks, this model places particular emphasis on the governance of digital services and their compatibility with national policy objectives. Nevertheless, China should not be understood as having a single, uniform regulatory instrument: its approach consists of multiple laws, administrative measures, and sector-specific rules. The comparison with the EU and the United States consequently reveals not three completely isolated systems, but different priorities concerning the allocation of authority among governments, businesses, and individuals.

These differences become especially problematic when AI systems operate across jurisdictions. A system that complies with the requirements of one country may encounter conflicting obligations elsewhere concerning data transfers, transparency, content moderation, risk assessment, or the disclosure of information about its operation. Multinational developers must navigate these differences, while smaller companies may lack the financial and legal resources necessary to comply with multiple regulatory regimes. Regulatory fragmentation can therefore create barriers to market entry, encourage the concentration of AI development among large corporations, and make international cooperation more difficult. At the same time, harmonization cannot simply mean imposing one country's regulatory preferences on the rest of the world. Legal standards reflect political choices and social values, and efforts to universalize a particular model may themselves become instruments of geopolitical competition.

The limitations of international legal coordination are particularly visible in the distinction between binding obligations and voluntary commitments. Initiatives developed through the United Nations, UNESCO, the OECD, and other international institutions have contributed to the articulation of principles concerning human rights, transparency, accountability, and trustworthy AI. UNESCO's Recommendation on the Ethics of Artificial Intelligence, adopted in 2021, provides a global normative reference, while the Council of Europe's Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, opened for signature in 2024, represents an important treaty-based development. However, these instruments differ in legal status, scope, participation, and implementation mechanisms. Moreover, the participation of states in international initiatives does not automatically ensure consistent domestic enforcement or agreement on the practical meaning of common principles [3, p. 42].

The central difficulty is that states are being asked to cooperate on a technology that they simultaneously regard as a source of strategic advantage. Governments may recognize the need to reduce risks associated with AI while remaining reluctant to accept international obligations that could constrain military applications, industrial policies, intelligence capabilities, or domestic information governance. This produces a structural dilemma: the greater the perceived strategic value of AI, the stronger the incentive to preserve national discretion; yet the more states prioritize unilateral control, the harder it becomes to address risks that transcend national boundaries. International coordination is therefore constrained not only by differences in legal doctrine but also by the distribution of power within the international system.

A viable approach to this problem requires neither complete regulatory uniformity nor unrestricted national discretion. Instead, international cooperation should concentrate on establishing a common minimum of safeguards while allowing states sufficient flexibility to adapt implementation to their legal systems and institutional capacities. Priority areas include the protection of fundamental rights, mechanisms for reporting serious AI-related incidents, cooperation on cross-border investigations, shared technical standards, and clearer rules concerning responsibility for transnational harm. Such arrangements would not eliminate geopolitical competition, but they could reduce the risk that regulatory differences become a source of instability or that states exploit legal gaps to avoid accountability.

Ultimately, the fragmentation of AI governance reflects a broader crisis in the capacity of international law to respond to technological developments that evolve faster than traditional processes of treaty negotiation and institutional adaptation. The challenge is not simply to create more rules, but to establish mechanisms through which competing regulatory systems can interact without undermining essential protections. The future effectiveness of AI governance will depend on whether states can recognize that strategic competition and international cooperation are not necessarily mutually exclusive. The decisive question is whether AI regulation will develop into a system of competing technological blocs or become a framework in which geopolitical differences are managed through shared legal commitments. The answer will shape not only the future of AI but also the credibility of international law as an instrument for governing emerging technologies.

The accountability gap: legal responsibility for ai-driven decisions and transboundary harm. One of the most challenging problems in the legal regulation of artificial intelligence is the growing discrepancy between the capacity of AI systems to influence consequential decisions and the ability of existing legal frameworks to determine responsibility for their outcomes. Artificial intelligence is increasingly used in areas such as healthcare, financial services, public administration, criminal justice, and military operations, where errors may result in substantial economic losses, violations of fundamental rights, or threats to human life. Yet when an AI-driven decision causes harm, identifying the legally responsible party may prove considerably more difficult than in conventional cases involving human conduct. Developers design the systems, companies integrate them into products, institutions determine how they are used, and individual operators may rely on their recommendations. Responsibility is consequently distributed across a complex network of actors, creating what may be described as an accountability gap: a situation in which harm is identifiable, but the applicable legal rules do not readily establish who should answer for it and on what grounds [4, p.69].

Traditional legal systems generally rely on concepts such as intention, negligence, foreseeability, causation, and the existence of a duty of care. Although these concepts remain relevant to AI-related disputes, their application becomes more complicated when decisions emerge from systems whose internal operations are difficult to interpret. Machine-learning models may identify statistical relationships that were not explicitly anticipated by their developers, while their outputs can change depending on the data, operational environment, and interactions with other systems. This does not mean that AI decisions are inherently unpredictable or that legal responsibility becomes impossible to establish. Rather, it challenges conventional assumptions about how conduct, causation, and fault can be demonstrated. A court may need to determine whether harm resulted from defective design, inadequate training data, insufficient testing, negligent deployment, improper human supervision, or a combination of these factors. Establishing the relevant causal relationship may require technical evidence that neither the injured party nor the court can easily access.

The problem becomes particularly acute when the parties involved occupy different positions within the AI development and deployment process. A developer may argue that its model was used in an unintended context; a deploying organization may claim that it relied on a commercially available system; and an operator may maintain that the system's recommendation appeared technically reliable. Each party may possess only partial knowledge of the circumstances that produced the harmful outcome. In addition, contractual arrangements may distribute operational duties among several organizations without necessarily resolving their obligations toward injured third parties. The result is a risk that the complexity of the technological ecosystem will become a practical obstacle to obtaining compensation or establishing accountability. An effective legal framework must therefore distinguish between the allocation of responsibility among commercial actors and the protection of those who suffer harm.

A related difficulty concerns the relationship between human oversight and meaningful human responsibility. The mere presence of a human decision-maker does not necessarily guarantee effective control over an AI system. In organizations where automated recommendations are produced rapidly and repeatedly, employees may gradually develop excessive confidence in algorithmic outputs, particularly when they lack the technical expertise or institutional authority to challenge them. This phenomenon, often associated with automation bias, can transform nominal human supervision into a procedural formality. Conversely, assigning responsibility exclusively to an individual operator may be unjustified when the person had insufficient information, inadequate training, or no realistic opportunity to override the system. The legal significance of human oversight should therefore be assessed by examining whether the person possessed the competence, information, authority, and practical opportunity necessary to intervene.

These challenges are reflected in contemporary regulatory developments. The European Union's Artificial Intelligence Act establishes obligations for different categories of AI systems and relevant economic operators, including requirements concerning risk management, documentation, human oversight, and monitoring in specified circumstances. However, regulatory compliance and civil liability address different questions: the fact that an organization has complied with applicable regulatory requirements does not automatically resolve every dispute concerning damage, causation, or compensation. Likewise, the existence of a regulatory violation does not eliminate the need to establish the relevant legal consequences under the applicable liability regime. Effective governance therefore requires coordination between preventive regulation, civil liability, procedural rules, and mechanisms through which injured parties can obtain evidence [7, p. 208].

The accountability problem becomes even more complex when AI-related harm crosses national borders. Consider a hypothetical case in which a model developed in one country is supplied by a company incorporated in another, hosted on infrastructure located in a third jurisdiction, and used by a public institution in a fourth. If the system unlawfully discriminates against individuals or causes significant financial damage, several legal questions arise simultaneously: which court has jurisdiction, which country's law governs the dispute, which party must disclose relevant technical evidence, and whether a judgment can be enforced against an organization located abroad. Differences in national liability standards and procedural safeguards may create uncertainty for victims while increasing the costs of litigation. In some circumstances, the injured party may face a substantial imbalance of resources and technical knowledge compared with the organizations responsible for developing and operating the system.

Cross-border harm also extends beyond conventional private disputes. AI-generated disinformation may affect elections in another state, automated systems may contribute to discriminatory decisions involving foreign nationals, and AI-enabled cyber operations may produce consequences across multiple jurisdictions. Where the conduct involves states or state-supported actors, questions of international responsibility may arise alongside domestic civil or criminal liability. Nevertheless, international responsibility should not be confused with the liability of private developers or users: the applicable legal framework depends on the identity of the actor, the nature of the conduct, and the relevant international obligations. Nor does the use of AI automatically create a separate category of international legal responsibility. Existing rules may remain applicable, although their interpretation and practical implementation can become more difficult when technological systems mediate the conduct in question.

Addressing this accountability gap requires a combination of legal clarification and institutional reform. First, responsibilities should be allocated according to the actual functions, knowledge, and control exercised by developers, providers, deployers, and operators, rather than being assigned exclusively to the actor closest to the harmful outcome. Second, procedural rules should ensure that affected individuals have realistic opportunities to obtain relevant evidence, particularly where essential technical information is controlled by the organizations involved. Third, risk-based insurance and compensation mechanisms may help ensure that victims are not left without an effective remedy when conventional litigation proves excessively difficult. Finally, greater international cooperation is necessary to facilitate evidence sharing, clarify jurisdictional questions, and improve the enforcement of decisions involving transnational AI services.

Ultimately, the accountability gap exposes a fundamental principle of the relationship between technological innovation and the rule of law: the complexity of a technological system must not become a justification for the absence of legal responsibility. AI may complicate the identification of fault, the reconstruction of causation, and the allocation of obligations, but these difficulties do not establish that responsibility should disappear. On the contrary, the more extensively AI systems influence decisions affecting human rights, public safety, and international stability, the more important it becomes to ensure that every consequential deployment remains embedded within a structure of identifiable duties and effective remedies. The objective of legal regulation should not be to attribute human intentions to machines, but to ensure that the individuals and institutions designing, controlling, and benefiting from AI remain answerable for the consequences of its use [8, p. 52].

AI, human rights, and democratic governance: between technological efficiency and political control. Artificial intelligence is increasingly reshaping the relationship between the state and the individual, creating new opportunities for public administration while raising fundamental questions about human rights, democratic accountability, and the legitimate exercise of political power. Governments can use AI to improve public services, detect fraud, allocate resources, and respond more efficiently to social needs. However, the same technological capabilities can also facilitate mass surveillance, influence political behaviour, restrict access to information, and automate decisions that affect individuals' rights and opportunities. The central political and legal dilemma is therefore not whether AI can make governance more efficient, but under what conditions technological efficiency remains compatible with human dignity, individual autonomy, and democratic control.One of the most significant concerns is the expansion of surveillance capabilities. AI enables authorities and private organizations to process enormous quantities of information, identify patterns of behaviour, and connect data collected from different sources. Facial recognition, biometric identification, predictive analytics, and automated monitoring can support legitimate objectives, including the investigation of serious crimes and the protection of public safety. Yet their deployment may also weaken the practical boundaries between lawful observation and intrusive supervision. When individuals believe that their movements, communications, and associations are continuously monitored, they may begin to modify their behaviour even in the absence of direct coercion. This chilling effect can undermine freedom of expression, peaceful assembly, and political participation. The resulting danger is not limited to unlawful data collection; it also concerns the gradual normalization of a political environment in which the exercise of fundamental freedoms becomes conditioned by the possibility of continuous observation.

The protection of privacy in the age of AI consequently extends beyond the confidentiality of personal information. It concerns an individual's ability to maintain a sphere of autonomy free from unjustified profiling, prediction, and manipulation. AI systems may infer sensitive characteristics from apparently ordinary data, classify individuals according to probabilistic assessments, or generate predictions about their future behaviour. Such inferences may influence decisions concerning employment, credit, insurance, education, and access to public services. Even when a system does not explicitly use a protected characteristic, its outputs may reproduce existing inequalities through indirect indicators or historically biased datasets. This creates a difficult legal question: should a decision be considered acceptable merely because it was generated through statistical analysis rather than an explicit discriminatory intention? From a human-rights perspective, the answer must depend on the decision's actual effects, its justification, and the availability of effective safeguards, rather than solely on the apparent neutrality of the technology [6, p.245].

The implications for democratic governance are equally significant. Democratic systems depend on citizens' ability to access reliable information, form independent political judgments, and participate in public debate without coercion or systematic manipulation. AI-generated content can support political communication by making information more accessible and enabling rapid responses to public concerns. At the same time, it can be used to produce deceptive images, fabricated statements, synthetic recordings, and personalized messages designed to exploit individual preferences or vulnerabilities. The danger becomes particularly serious when these techniques are deployed during elections, periods of political instability, or international crises. Unlike conventional propaganda, AI-assisted influence operations can be automated, rapidly modified, and directed toward narrowly defined audiences. Their effectiveness may lie not only in persuading citizens to accept a particular narrative, but also in generating confusion, deepening polarization, and undermining confidence in democratic institutions.

Nevertheless, the relationship between AI and democracy should not be reduced to a simple opposition between technology and freedom. AI can strengthen democratic institutions by improving access to government information, assisting public consultations, identifying patterns of corruption, and making administrative processes more transparent. The decisive factor is the institutional environment in which these capabilities are deployed. When automated systems operate under clear legal mandates, independent oversight, meaningful transparency, and accessible complaint mechanisms, they may improve the quality of public governance. When they operate without adequate scrutiny, the same systems may concentrate decision-making power and make public institutions less responsive to citizens. Technological efficiency is therefore not an independent measure of democratic progress: it must be evaluated alongside legality, fairness, accountability, and the preservation of meaningful human judgment.

A particularly sensitive issue concerns the use of AI in public administration and law enforcement. Automated systems may assist authorities in identifying irregularities, prioritizing inspections, or allocating limited resources. However, when historical data reflect discriminatory practices or unequal patterns of enforcement, algorithmic recommendations may reproduce these patterns under the appearance of objective analysis. Furthermore, decisions generated by complex models can be difficult for affected individuals to challenge if the reasons behind them are not adequately explained. The right to an effective remedy becomes particularly important when automated assessments influence access to social benefits, immigration procedures, criminal justice, or other areas in which state decisions can profoundly affect an individual's life. Human oversight in these contexts must therefore be substantive rather than symbolic, ensuring that responsible officials can examine the grounds of a decision, question its reliability, and correct errors.

These concerns also reveal a broader tension between national security and fundamental rights. Governments may argue that extensive data analysis and automated monitoring are necessary to address terrorism, cyber threats, foreign interference, or organized crime. Such objectives can be legitimate, but they do not automatically justify every technological measure adopted in their name. Restrictions on fundamental rights must be assessed under the applicable legal framework, including requirements of legality, legitimate purpose, necessity, and proportionality. The availability of powerful analytical tools does not eliminate the obligation to demonstrate that their use is justified. Indeed, the capacity of AI to process information on an unprecedented scale makes independent supervision especially important, since the cumulative effects of multiple surveillance and profiling systems may be difficult to assess when each is considered in isolation [6, p. 211].

The international dimension of these challenges is becoming increasingly important. Divergent approaches to privacy, freedom of expression, online content, and state surveillance complicate efforts to establish universally accepted safeguards. Some governments emphasize individual rights and independent oversight, while others prioritize state security, social stability, or centralized control over the information environment. These differences are reflected in international debates over the regulation of digital platforms and AI systems. Yet political disagreement does not remove the need for cooperation on common concerns, including discriminatory automated decisions, the protection of vulnerable populations, and the misuse of synthetic media. International human-rights standards provide an important reference point, although their effective application requires institutions capable of translating general principles into enforceable obligations and practical remedies [7, p. 113].

Ultimately, the legitimacy of AI-enabled governance depends on whether technological capabilities remain subject to meaningful democratic and legal control. Efficiency may justify the adoption of automated tools, but it cannot by itself justify the transfer of consequential public decisions to systems that affected individuals cannot question or challenge. Nor should the preservation of human oversight be understood as a purely technical requirement: it expresses the deeper principle that public authority must remain answerable to the people whose rights it affects. The fundamental test of democratic AI governance is not how much decision-making can be automated, but whether automation strengthens the capacity of individuals to exercise their rights and hold institutions accountable. A regulatory framework that places human dignity, political freedom, and effective remedies at its centre can help ensure that AI serves as an instrument of democratic development rather than a mechanism for concentrating power.

1. Crawford K. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. – New Haven: Yale University Press, 2021. – 336 p.

2. Bostrom N. Superintelligence: Paths, Dangers, Strategies. – Oxford: Oxford University Press, 2014. – 352 p.

3. Russell S. Human Compatible: Artificial Intelligence and the Problem of Control. – New York: Viking, 2019. – 352 p.

4. Eubanks V. Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor. – New York: St. Martin’s Press, 2018. – 288 p.

5. O’Neil C. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. – New York: Crown, 2016. – 272 p.

6. Susskind R. A World Without Work: Technology, Automation, and How We Should Respond. – London: Allen Lane, 2020. – 352 p.

7. Zuboff S. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. – New York: PublicAffairs, 2019. – 704 p.

8. Nye J. S. Jr. The Future of Power. – New York: PublicAffairs, 2011. – 320 p.

Тарана ШАХБАЗОВА (Азербайджан),
 специалист в области международных отношений

Резюме. В статье исследуются политико-правовые проблемы регулирования искусственного интеллекта в контексте трансформации международных отношений и глобальной системы управления. Рассматриваются влияние искусственного интеллекта на государственный суверенитет и стратегическую конкуренцию, противоречия между национальными моделями регулирования, сложности установления юридической ответственности за решения, принимаемые с использованием ИИ, а также риски для прав человека и демократического управления. Особое внимание уделяется фрагментации международно-правовых механизмов и необходимости укрепления многостороннего сотрудничества. Обосновывается положение о том, что эффективное регулирование ИИ требует согласования национальных интересов с международными обязательствами, обеспечения прозрачности и подотчётности, а также сохранения содержательного человеческого контроля над решениями, затрагивающими основные права и свободы. Делается вывод о необходимости формирования многосторонней системы управления искусственным интеллектом, способной обеспечить согласование технологического развития с защитой фундаментальных прав и принципом международной ответственности.

Ключевые слова: искусственный интеллект, международные отношения, государственный суверенитет, глобальное управление, международное право, юридическая ответственность, права человека, цифровое неравенство, международное сотрудничество.