International Legal regulation of AI - Key Challenges and Perspectives

by

 Yevheniia Khoroshun

CyJurll Scholar 

on 30 July 2026

Introduction

The rapid pace of technological development, combined with globalization, has accelerated the cross-border spread of cutting-edge technologies, including technologies of Artificial Intelligence (hereinafter - AI). As these systems can be developed and deployed simultaneously across multiple jurisdictions, they give rise to new risks that extend beyond national borders. Consequently, the regulation of AI is not exclusively a domestic issue, but rather a pressing matter of international law. As UN Secretary-General António Guterres noted regarding the preliminary report of the Independent scientific panel on Artificial Intelligence: ‘The more AI advances without shared rules, the less say governments and people will have in the outcome’[1]. This warning underscores the urgent need for coordinated global legal measures.

Discussion

The pioneering nature of AI technologies and the global scale of their application is driving the development of entirely new international regulatory frameworks. At the same time, it presents unprecedented challenges and complex issues for lawmakers seeking to implement them.

The initial challenge to be addressed is the lack of universally recognized definition for AI. This definitional deficit generates ambiguity concerning the exact nature of the systems subject to legal scrutiny, thereby accelerating asymmetrical enforcement across national boundaries. It is undeniable that formulating a comprehensive definition is intrinsically challenging, given the multi-faceted modalities, diverse application domains and evolving characteristics of AI technologies. Yet, establishing a unified conceptual framework remains essential for providing a stable groundwork for upcoming international legislation.

Currently, distinct definitions of AI coexist within key legal frameworks, including the 2024 Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law [2];  EU Artificial Intelligence Act [3] and UNESCO Recommendation on the Ethics of Artificial Intelligence [4] (soft law). These documents establish a common baseline - specifically, identifying AI as a machine-based system that uses inputs to generate outputs like predictions, recommendations or decisions that affect real or virtual environments. However, they split when it comes to specific features, leaning either toward functional capabilities (like learning and cognitive processing in UNESCO`s Guidelines definition) or technical attributes (like autonomy and inference in EU AI Act and Council of Europe`s terminology). Consequently, this lack of a standardized definition leaves a gap of uncertainty that could lead to poor outcomes for international legal regulation.

The next complication is a wide range of applications of AI and the multifaceted nature of the challenges raised by it. The application of task-specific AI spans multiple domains, from scientific research to healthcare and education. These systems are adopted by different types of actors - private companies seeking business benefits, public authorities automating administrative processes, and individuals using AI for personal tasks [5]. The technologies themselves appear in a variety of forms and their organizational contexts differ widely [6]. Across these contexts, AI deployment generates a range of various challenges, such as risks to privacy, fairness, accountability, and fundamental human rights. Addressing these issues requires a set of high-level, adaptable principles and rules that are sufficiently general to cover many AI systems but specific enough to guide responsible design, implementation, and deployment.

A subsequent challenge is the extremely rapid development of AI technologies. It renders regulation of AI as a fixed or “static” construct impracticable, because the nature and magnitude of risks continuously evolve. 2026 findings from an Independent International Scientific Panel on AI indicate that the evaluation of AI behaviour is challenged by features like memorization of training outputs and active deceptive behaviour [5]. These characteristics make robust risk assessment problematic, thereby complicating the evidence base required for effective law-making. The lack of a reliable characterization of risks creates a serious impasse when it comes to establishing clear restrictions, thresholds and permissible uses of AI in various contexts.

Compounding these difficulties is the further challenge - the fragmentation of the existing regulatory framework. Regulatory divergence stems from disparities in economic resources and from differing policy priorities shaped by national economic interests, cultural values and technological capacities, which may significantly hinder international cooperation relating to AI technologies. Reliance solely on regional legal frameworks can also prove insufficient and may produce adverse effects, as AI functions globally [7].

Regional approaches vary. Binding regional instruments - such as the EU Artificial Intelligence Act [3] and the Council of Europe’s Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law [2] - adopt goal-oriented frameworks that regulate the conduct of principal actors in the AI ecosystem, including states, developers, and end users. Complementary regional soft-law instruments, for instance ASEAN Guide on AI Governance and Ethics [8] - also offer normative and conceptual guidance for interpreting these issues, but they remain non-binding and primarily recommendatory.

National regulations are what makes the regulating landscape even more uneven. While certain emerging frameworks, such as Canada’s Artificial Intelligence and Data Act (AIDA) [9], seek alignment with evolving international norms, domestic laws must inevitably interface with complex, pre-existing statutory ecosystems. For instance, China’s Interim Measures for the Management of Generative Artificial Intelligence Services applies extraterritorially to all public-facing providers within its jurisdiction, functioning concurrently with established frameworks like the Cybersecurity Law of the PRC, the Personal Information Protection Law of the PRC and various sector-specific guidelines [10]. Furthermore, national legislation does not always govern AI holistically. Instead, it often targets domain-specific applications, as evidenced by the Law of Ukraine on Academic Integrity, regulating AI integration within academic conduct policies [11]. Cumulatively, these localized interventions risk fostering a highly fragmented legal ecosystem that will prove increasingly difficult to harmonize in the future.

To navigate the challenges outlined above, a multifaceted governance strategy is required, structured around global harmonization, sector-specific adaptation and repetitive legislative review.

With regard to general regulation, the aim should be an establishment of common rules and principles on issues relating to AI.  Preference should be given to a modular treaty framework, with multiple treaties each addressing specific issues regarding AI, rather than a single rigid model. Specialized treaties may regulate the development of AI, whilst separate legal instruments might govern its use by different actors, for example by public authorities, commercial entities and individual users. This framework should be accompanied by a harmonised, uniform definition of the term ‘AI’. 

For subject‑specific concerns, new international instruments should be adopted to regulate AI applications in particular areas and existing international treaties should be updated, for example through protocols or annexes, to incorporate AI‑relevant provisions.

To ensure that international regulation remains aligned with technological advances, international legal standards should be subject to regular review with a mandate to amend and strengthen instruments that prove inadequate in addressing emergent AI risks. Continuous review mechanisms will enable timely adjustments and greater legal resilience in a rapidly evolving field.

Conclusion

The current international regulation of AI remains challenged by a lack of a standardized definition, the rapid and unpredictable evolution of AI capabilities and a disjointed ecosystem of regional and sector-specific laws. Left unaddressed, these regulatory gaps risk creating loopholes and inconsistent enforcement that could undermine fundamental human rights and international security.

To prevent a ‘shortage of definitions’ and legislative obsolescence, a modular system of treaties based on a single definition of AI could be introduced. To ensure the resilience of international law in the face of relentless technological progress, thematic protocols could be added to existing instruments, and mechanisms for the ongoing review of legislation should be established. Ultimately, only a harmonized and highly adaptable legal response can ensure that AI serves as a tool for global progress rather than a source of instability.

References

1)              United Nations, Secretary-General, 'Press Conference with the Co-Chairs of the Independent Scientific Panel on Artificial Intelligence' (press conference, New York, 1 July 2026). <https://www.un.org/independent-international-scientific-panel-ai/sites/default/files/2026-07/secretary_general_remarks_-_ai_scientific_panel_preliminary_report_launch.pdf>   [accessed 05 July 2026]

2)              Council of Europe: Committee of Ministers, Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, CETS No. 25, [17 May 2024], <https://www.refworld.org/legal/agreements/coeministers/2024/148016>  [accessed 10 July 2026]

3)              Regulation (EU) 2024/1689 of the European Parliamentand of the Council of 13 June 2024 (Artificial Intelligence Act) [2024] OJ 2 144/1 , <https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng > [accessed 06 July 2026]

4)              UNESCO, Recommendation on the Ethics of Artificial Intelligence, SHS/BIO/PI/2021/1 [2022] <https://unesdoc.unesco.org/ark:/48223/pf0000381137_eng> [accessed 05.07/2026]  

5)              Independent International Scientific Panel on AI, Preliminary Report of the Independent International Scientific Panel on AI: Evidence-based assessment of opportunities, risks and impacts of artificial intelligence (July 2026) <https://www.un.org/independent-international-scientific-panel-ai/sites/default/files/2026-07/en_Preliminary%20Report_.pdf > [accessed 7 July 2026].

6)              International Science Council [September 2025]. Types of AI and their use in Science. DOI: 10.24948/2025.09   [accessed 10 July 2026]

7)              Olena Chernenko, 'Problems of legal regulation artificial intelligence technology' [2024] 1(24) Collection of Scientific Papers «Private Law and Business», <https://doi.org/10.32849/2409-9201.2024.24.14> [accessed 07 July 2026]

8)              Association of Southeast Asian Nations, ASEAN Guide on AI Governance and Ethics (February 2024) <https://asean.org/wp-content/uploads/2024/02/ASEAN-Guide-on-AI-Governance-and-Ethics_beautified_201223_v2.pdf > [accessed 05 July 2026]

9)              Government of Canada, Innovation, Science and Economic Development Canada, The Artificial Intelligence and Data Act (AIDA) – Companion Document (6 April 2026) <https://ised-isde.canada.ca/site/innovation-better-canada/en/artificial-intelligence-and-data-act-aida-companion-document#s5 > [accessed 07 July 2026]

10)          People’s Republic of China, Interim Measures for the Management of Generative Artificial Intelligence Services (10 July 2023) <http://www.cac.gov.cn/2023-07/13/c_1690898327029107.htm > [accessed 08 July 2026] 

11)          Ukraine, Law of Ukraine on Academic Integrity (No 4742-IX, 18 December 2025) <https://zakon.rada.gov.ua/laws/show/4742-20#Text > [accessed 10 July 2026]