ABSTRACT

1 INTRODUCTION

As artificial intelligence (AI) advances and its applications expand, its significance continues to grow, exerting an increasing impact on society and shaping future development.¹ The majority of economic sectors, social interactions, and technical breakthroughs are expected to depend on AI as a foundational technology. AI is fast evolving with the potential to improve business operations, enhance public safety, and contribute to broader social progress.² At the same time, there will be issues, some foreseen, and many that will evolve alongside the technology itself. Regulating AI through traditional governance systems is challenging due to its pervasive and evolving nature. Instead, a degree of flexibility to promote innovation while ensuring security has often been provided through a variety of "soft-law," or non-binding, instruments.

Statutory law, which is the term used for written laws passed by a legislative body, is intended to give people and organisations a framework within which it is supposed to operate. Yet the regulation of AI's development and application has become increasingly complex due to society’s growing reliance.³

Statutory law has long served as the cornerstone of legal systems, but in the era of artificial intelligence, these shortcomings have come to attention.⁴ This research will examine the limitations of statutory law in the context of AI and explore how technological advances are disrupting established legal norms. It is becoming clear that statutory law alone cannot adequately handle the ethical, societal, and economic consequences of AI.

The widespread deployment of AI across diverse industries poses pressing issues of control and governance.⁵ The rapid development of AI systems frequently surpasses the creation of related legal frameworks, posing significant challenges for legislators. Efforts to establish accountability and transparency are hindered by the dynamic nature of AI algorithms, which evolve and adapt within decision-making processes.⁶ Regulatory efforts are further complicated by technical challenges, including bias and interpretability.

This research seeks to explore how statutory law can be adapted to effectively regulate artificial intelligence, striking a balance between fostering technological innovation and ensuring societal responsibility, while acknowledging its inherent limitations in addressing AI's unique ethical, social, and economic challenges. 2 METHODOLOGY AND RESEARCH APPROACH

2 METHODOLOGY AND RESEARCH APPROACH

This study examines the role of statutory law in AI regulation across several jurisdictions, employing a comparative legal analysis methodology in conjunction with doctrinal research methods. To explore how various legal systems address AI governance issues, the research employs a qualitative research methodology, combining a systematic examination of legislative texts, regulatory proposals, and court rulings.

The methodological framework consists of a few essential elements. First, doctrinal legal research forms the foundation of the analysis, involving a thorough examination of primary legal sources such as laws, rules, and case law pertaining to AI governance. This method

reveals weaknesses in existing regulatory frameworks and permits a methodical analysis of how current legal frameworks respond to emerging AI challenges. Comparative legal analysis, focusing on the European Union's AI Act in conjunction with the latest frameworks in the US, UK, Brazil, and China, offers insights into various regulatory philosophies and approaches across jurisdictions.

The research methodology also incorporates analysis of secondary sources, including academic literature, policy documents, and industry reports, to provide contextual understanding of regulatory challenges and opportunities. This multi-source approach ensures comprehensive coverage of both theoretical frameworks and practical implementation challenges in AI regulation. The IRAC method (Issue, Rule, Application, and Conclusion) is used as an analytical framework to structure the legal analysis and ensure systematic examination of AI regulatory challenges.

Data collection was conducted through a systematic review of legislative texts, regulatory proposals, and academic commentary published between 2018 and 2024. This temporal scope was selected to encompass the most significant period of AI regulatory development globally, beginning with the European Union's initial AI regulatory proposals and extending to recent legislative initiatives in multiple jurisdictions.

3 HISTORICAL CONTEXT AND LEGISLATIVE EVOLUTION

Throughout history, the development of AI has been influenced by broader socio-political and economic environments, which in turn have shaped public attitudes and governmental responses. Early debates surrounding AI centred on how society would change as a result of it; however, concerns about job displacement and ethical dilemmas soon emerged, particularly in relation to productivity gains.

Changes in the availability of legal information have had an impact on the evolution of the legal services delivery system. With the invention of the CD-ROM in the 1990s, the use of digital resources and libraries gained momentum in the process of evolution.⁷ Today, nearly all law firms rely on digital legal materials.⁸ Predictive algorithms represent the most recent stage of this evolution, enabling lawyers to navigate complex legal dilemmas and identify and synthesise relevant information.

Computer-assisted legal research, pioneered by Westlaw and Lexis in 1976, has become a cornerstone of legal practice. Most legal research is now conducted online,⁹ replacing reliance on physical libraries. Digital access to legal texts allows attorneys to locate pertinent sources through keyword searches, significantly reducing the time once required to manually consult indices and read through each source individually.¹⁰ While larger institutions rely heavily on proprietary data providers such as Westlaw, Lexis, or Bloomberg, others turn to publicly available sources.¹¹

Traditionally, lawyers have been regarded as highly competent individuals trained to identify the relevant facts, frame pertinent legal issues, and predict the likely outcome of the case.¹² Attorneys apply judgment to evaluate the merits of a case and determine the best course of action by drawing on expertise and intuition. For a long time, such tasks were considered the exclusive domain of highly qualified specialists/ lawyers.

However, modern advances in AI have challenged long-held beliefs about human knowledge, particularly in the areas of machine learning and natural language processing. It is clear that the impact of data-driven analysis extends to the practice of law.¹³ Litigation itself is gradually evolving, with disputes increasingly being resolved “in the shadow of the law”, where settlement outcomes are shaped by the likely decision a court would reach. Predictions made by algorithms are repeatable by others.

The legislative development of AI can be traced through a series of significant events in several countries. Early regulations sought to advance R&D while maintaining ethical and safety standards. As AI applications spread across several industries, regulators faced new challenges concerning algorithmic transparency, cybersecurity, and data privacy.¹⁴ In response, legislative solutions have been progressively updated to consider new dangers and public concerns. The definition of AI itself has remained fluid, shifting with technological advances.¹⁵ In recent years, there has been a growing number of measures

to improve algorithmic accountability and transparency, including inquiries for regulatory monitoring of AI systems, particularly in high-stakes sectors such as criminal justice, finance, and healthcare.¹⁶

China's AI regulatory framework emphasises enhancing norms for scientific and technological ethics, focusing on data security protection systems and balancing fair competition with innovation encouragement. It addresses ethical concerns, attribution of liability, and the prevention of intellectual property monopolies. A central component is the Interim Administrative Measures for Generative Artificial Intelligence Services,¹⁷ which plays a key role in ensuring governance mechanisms work together to build trust and accountability in the AGI industry.¹⁸

Nevertheless, China’s current AI regulatory framework remains limited in scope, particularly in its lack of comprehensive definitions and protections for data rights. While it underscores the need for specialised intellectual property protection for data and the adoption of anti-monopoly measures to prevent misuse and monopolistic practices, it does not provide a fully developed legal framework. A more legal framework—one that recognises data as a form of intellectual property, acknowledges its dual public and proprietary nature, and strengthens digital IP protections alongside ethical guidelines for AI—is still needed.¹⁹

Brazil's approach is primarily outlined in Bill 21/2020,²⁰ which sets out the objectives and foundational principles for the development and use of AI, including ethical considerations. However, the bill does not explicitly regulate data privacy or intellectual property. Critics argue that its abstract provisions and limited number of articles risk overlooking essential aspects such as data protection and ethical guidelines, necessitating further discussion and refinement in the Senate.²¹

While previous sections outlined individual jurisdictional approaches, a systematic comparison reveals fundamental divergences in how major economies balance innovation and responsibility through statutory frameworks:

The EU's AI Act (2024)²² exemplifies a risk-based hierarchical model that prohibits unacceptable practices, such as social scoring and imposes stringent ex-ante requirements for high-risk systems, including conformity assessments and fundamental rights impact evaluations. Unlike the US's sectoral approach, the EU centralises enforcement through a European AI Office, creating uniform compliance burdens. Critics argue this may stifle startups lacking resources for compliance,²³ while proponents highlight its strong emphasis on safeguarding fundamental rights.

By contrast, the US regulates AI through fragmented sectoral agencies such as the FTC, FDA, and NTSB, relying largely on non-binding frameworks like the NIST AI RMF and the Blueprint for an AI Bill of Rights. Proposed legislation, such as the Algorithmic Accountability Act (2023)²⁴, focuses narrowly on impact assessments in specific contexts such as hiring and housing. This avoids EU-style centralised burdens but creates regulatory uncertainty and enforcement gaps. Further, state-level initiatives—such as California’s AB 331—push stricter rules, risking a fragmented "patchwork" regime. The absence of federal AI liability laws leaves accountability reliant on tort law, creating ambiguity for autonomous systems.

China’s approach merges aggressive state investment in AI R&D with strict control mechanisms, exemplified by the Generative AI Interim Measures (2023). Its "negative list" system prohibits challenges to state authority while promoting industrial dominance in non-sensitive sectors. Unlike Western models that prioritise individual rights, China emphasises data sovereignty, social stability, and alignment with "socialist core values." This facilitates rapid scaling of state-approved innovations but restricts algorithmic transparency and independent oversight.

Brazil’s Bill 21/2020 mirrors EU principles (human oversight, non-discrimination) but lacks implementation mechanisms. The bill coexists with the LGPD (GDPR-inspired data law), creating potential conflicts between data minimisation and AI training needs. More broadly,

emerging economies often adopt EU-style principles but encounter capacity gaps in enforcement. In several cases, such as India’s Digital India Act, governments prioritise "sovereign AI" infrastructure over ethics frameworks, reflecting divergent innovationresponsibility weightings.

4 ROLE OF STATUTORY LAW IN REGULATING ARTIFICIAL INTELLIGENCE

International cooperation in AI governance increasingly centres on initiatives to establish technological and ethical standards that ensure the responsible development and application of AI systems.²⁵ Organisations, including the International Telecommunication Union, the International Organisation for Standardisation, and the Organisation for Economic Co-operation and Development, have undertaken initiatives to create AI guidelines, standards, and principles that support accountability, openness, equity, and user-centred design.

Within this global landscape, statutory law outlines rules for the creation, implementation, and use of AI systems. It provides a legal framework within which individuals and organisations operate and serves several key regulatory functions.²⁶ First, clear guidelines and norms for the creation and use of AI systems are one of the key functions of statutory law in regulating the emerging technologies.²⁷ To ensure that AI systems do not endanger the public's health or safety, statutory law can set minimal safety criteria.²⁸ To further ensure that personal information and civil liberties are recognised, it can also establish rules for the gathering and use of data in AI systems.²⁹

The function of statutory law includes regulating AI systems by emphasising their liability and accountability.³⁰ If an AI system affects someone, the creator, the user, or the AI system

itself could all be held accountable according to statutory law.³¹ This is particularly crucial when AI systems are applied in high-risk industries, such as healthcare or transportation.³²

Statutory law also contributes to limiting the impact of AI on employment and the economy. To prevent discrimination, statutory legislation should explore rules for the application of AI to hiring and promotion decision-making processes. Additionally, it can establish rules for the application of AI in fields where job displacement is a concern.

4.1. Statutory Laws and Their Limitations

The limitations of statutory law in addressing challenges posed by AI have come to light. As AI systems grow in complexity and autonomy, their regulation becomes more difficult.³³ It might equally be challenging for regulators to comprehend how they operate or recognise possible concerns.³⁴ Considering how quickly AI technology is developing, it is challenging for statutory law to keep up with and adjust to new advances.³⁵

In principle, rules should be relied upon to regulate human conduct, as the law must strike a balance between flexibility, certainty, and reliability. Compared to principles, rules are more definite and easier to apply consistently. Law remains one of society’s most crucial instruments for shaping behaviour, offering rewards for certain actions and penalties for others, influencing the creation of social institutions.

However, the law can also distort individual decision-making. While awareness of legal consequences can be helpful, it may constrain a person's ability to act in a way that reflects their genuine preferences, moral convictions, and economic interests. An overemphasis on legal compliance risks fostering excessive strategic thinking and manipulative conduct. Ultimately, this is damaging to psychological well-being, distributive fairness, autonomy, and efficiency. Achieving an optimal equilibrium between the law’s beneficial function and its potential to distort behaviour is a challenging matter.

Insolvency law provides an example where concealment of legal rules is sometimes justified due to ex-ante strains.³⁶ When taking out a loan and investing its proceeds,

debtors bear full responsibility for repayment obligations.³⁷ However, if unable to repay, the law offers relief through various legal solutions to the creditor. If debtors were fully aware of these remedies beforehand, their behaviour might become more opportunistic or uncertain. In such cases, shielding the specifics of ex post legal remedies can be justified to preserve responsibility and fairness.

There can never be a definitive settlement of the relative jurisdictions of legislators and courts. Like other aspects of political organisation, this relationship is open to ongoing interpretation and discussion. In substantive criminal law, three principles govern the relationship between legislatures and courts: the legality principle, or nulla poena sine lege;³⁸ the constitutional theory of void-for-vagueness;³⁹ and the principle of strict construction, which dictates that courts must always construe criminal statutes in a way that favours the accused when there is any remaining ambiguity. Taken as a whole, these principles reflect a cautious approach to judicial lawmaking, limiting the judiciary’s role in the creation of criminal law.

At the same time, AI underscores the importance of aligning legal systems with social considerations and public policy goals.⁴⁰ The societal influence of AI is intricate and calls for a wide range of public policy solutions, from employment prospects to ethical issues. One major concern is the disruption of labour markets, which has sparked conversations about workforce transition plans, universal basic income, and retraining initiatives. Equity and fairness are often compromised by AI-driven algorithms, leading to issues of prejudice, discrimination, and unequal access to opportunities.

The formulation of laws regulating AI is complicated by the inherent restrictions imposed by statutory law.⁴¹ While statutory law is intended to give people and organisations a framework within which to operate,⁴² it may not always be able to foresee the particular difficulties brought on by innovative and quickly developing AI technology. Therefore, legal loopholes frequently emerge, undermining the effectiveness of statutory regulation and necessitating supplementary reforms.⁴³

The rise of AI presents a variety of challenges to conventional legal systems.⁴⁴ AI is increasingly applied in decision-making processes such as personnel selection, credit assessment, and aspects of legal analysis. However, these applications raise serious concerns about accountability, transparency, and fairness.⁴⁵ More innovative applications—such as driverless cars, AI-powered medical diagnostics, and autonomous drones—introduce regulatory questions that conventional legislation might not be able to handle as AI technologies advance and diversify. For example, the deployment of autonomous vehicles highlights unresolved issues regarding responsibility, safety requirements, and regulatory oversight, since current transportation laws do not adequately account for their unique capabilities and difficulties.

The broad application of AI in a variety of industries, such as healthcare, banking, and criminal justice, highlights the necessity of industry-specific laws designed to handle risks and issues unique to each industry.⁴⁶ However, the fragmented nature of regulatory initiatives, combined with the rapid pace of technological advancement, can lead to gaps and inconsistencies that undermine effective enforcement. Moreover, the global nature of AI development and application complicates governance:⁴⁷ diverging national frameworks and standards risk creating obstacles to innovation and interoperability. This makes international cooperation and harmonising regulatory frameworks essential for closing legal loopholes and ensuring accountability, safety and fairness in the use of AI technologies.

A fundamental limitation of statutory law is its reactive character. Statutory laws typically emerge in response to an existing issue or problem, which makes them resistant to adapting proactively to new developments. AI presents particular challenges in this regard: as AI evolves rapidly, new ethical and legal concerns continually emerge, often faster than legislatures can respond. Consequently, statutory law risks lagging behind technological change, leaving legal systems unprepared to address the unique issues brought on by the usage of AI.

The disadvantage of statutes is that they are often overly restrictive. Although statutory law is intended to be broadly applicable across a range of scenarios, this generality can make it difficult to address specific situations. This is particularly problematic for AI, which is routinely used in varied situations that do not neatly align with established legal frameworks.

Statutory law frequently faces limitations arising from its dependence on human interpretation. Because laws are expressed in natural language, they are inherently open to multiple interpretations.⁴⁸ This limitation is particularly salient in the context of AI, where algorithmic decisions can be difficult to understand or explain, potentially producing inconsistent and ambiguous legal outcomes. In State v. Loomis, the Wisconsin Supreme Court acknowledged the defendant's challenge to the use of the COMPAS risk assessment algorithm in sentencing. While upholding its use, the court mandated specific warnings for judges, highlighting judicial recognition of statutory law's struggle with algorithmic opacity and potential bias.⁴⁹ It may be challenging to prove that these biases exist, though, given the often inaccessible nature of AI decision-making processes.

The rigidity and constrained nature of statutory law are key factors limiting its ability to address the ethical and moral implications of AI.⁵⁰ Legal systems often struggle to keep pace with the remarkable rate at which AI technologies are developing,⁵¹ and the inflexible terminology of statutes may not take unforeseeable events or the broader ethical ramifications of AI into account.⁵²

Statutory law also relies heavily on human interpretation, which can be perplexing or unpredictable in situations involving intricate moral and ethical dilemmas.⁵³ Different interpreters may reach divergent conclusions, resulting in inconsistent or contradictory applications of the law.⁵⁴ Furthermore, statutory law typically concentrates on addressing certain problems, whereas AI has the potential to impact multiple facets of society.⁵⁵ This means that the profound ethical and moral consequences of AI may not be fully addressed.

The regulation of AI through statutory legislation faces a number of restrictions. Some of these restrictions include:

1. Lack of clarity: Terms associated with AI, such as machine learning algorithms, neural networks, and deep learning, often lack precise legal definitions, making it challenging to apply existing statutes to developing technology.
2. Slow legislative process: Enacting or amending statutes is typically a lengthy process. Consequently, statutory law often lags behind the rapid development of AI, creating regulatory gaps and loopholes.
3. Limited application: Statutory law often focuses on particular regulatory areas, such as liability, security, or privacy. This narrow focus may not be sufficient to address the complex and numerous problems that arise from the creation and application of AI.
4. Inflexibility: Once enacted, statutes can be difficult to amend or modify. This rigidity hinders timely responses to new and developing AI-related concerns, and the variation between national legal systems can complicate efforts to regulate AI on a global level.
5. Enforcement challenges: Enforcing statutory law can be challenging, particularly in the age of rapidly developing technologies like artificial intelligence. The complexity of AI systems, combined with ambiguities in existing legal provisions, can make it challenging to identify and hold violators accountable.

4.2. The Difficulties of Juggling Innovation and Responsibility

One of the central challenges in balancing innovation and accountability in AI governance is promoting the development of AI technologies while ensuring their use is responsible and ethical. ⁵⁶This entails addressing the statutory law's restrictions on the regulation of AI, the quick advancement of technology, and the broader moral, societal and economic implications of AI.

Three foundational arguments underlie key legal doctrines, including the vagueness theory, the rule of rigid construction, and nulla poena sine lege.⁵⁷ First, judicial innovation is rendered illegitimate when popular sovereignty is linked to legislative supremacy, reflecting the principle of "separation of powers" in modern constitutionalism. Second, it is unfair to penalise behaviour that was not previously classified as criminal, emphasising "notice" and "fair warning". Third, concerns about biased or arbitrary application of the criminal code highlight the importance of legal formalism in constraining unchecked discretion.

The so-called "rule of law" underlies both the vagueness theory and nulla poena sine lege. Yet there is danger in invoking this term, as it has become a highly malleable political catchphrase. Too often, the rule of law is conflated with the rule of good law, turning it into a blanket assertion of virtue within a legal framework. Legal theorists and philosophers,

however, have developed the idea in unduly complex ways. At its core, the prohibition of arbitrary behaviour in the use of state power is symbolised by the rule of law. In the context of criminal law, it requires that agents of official coercion, to the extent possible, act in accordance with established rules—that is, publicly recognised, reasonably stable, and broadly applicable declarations of prohibited behaviour by the state.

The economic impact of AI is another critical factor shaping industrial and societal development. AI integration can enhance productivity and foster innovation, but it also raises challenges pertaining to competitiveness, market dynamics, and regulatory control. Automation has the potential to transform the labour market, triggering issues such as job displacement, the need for skill retraining, and income inequality. At the same time, AI raises significant ethical and societal concerns, particularly when algorithms are used to make judgments in hiring, credit or criminal justice. Bias in algorithmic decision-making can have profound effects on people and communities.

The key considerations for AI governance are described below:

4.2.1. Privacy & Social Implications

Privacy is a critical area of concern that deserves attention while developing AI.⁵⁸ As machine learning algorithms evolve, they have the capacity to collect, analyse, and store vast amounts of personal data, raising questions about how such data is used and safeguarded.⁵⁹ This issue is particularly acute in healthcare, where AI is being applied to create innovative treatments and actions, creating concerns about the security and privacy of patient data.⁶⁰

The ethical governance of AI is further complicated by the limitations of statutory law. While statutory frameworks provide a structured environment in which people and organisations can function, they may be inadequate to address the unique ethical challenges posed by AI.⁶¹ Emerging AI systems can generate new types of harm that aren't protected by current legal frameworks, or they could cause concerns about the accountability and duty of those who utilise AI.⁶² The ability of AI technologies to make

choices and forecasts without human input introduces complex ethical dilemmas,⁶³ particularly related to privacy, bias, and discrimination.⁶⁴

AI systems often process extensive personal datasets, including biometric data, location data, and browser history, to generate predictions and judgments about individuals.⁶⁵ Without people's awareness or consent, this data can be gathered and utilised, raising privacy, security and monitoring concerns.

4.2.2. Bias and Discrimination Concerns

Machine learning algorithms can be trained on biased datasets, which may result in biased projections and conclusions, raising issues about prejudice and discrimination.⁶⁶ Such biases can disproportionately affect particular groups, including minorities or people with disabilities. Judicial awareness of the tension between technological tools and fairness is evident in cases such as State v. Loomis (Wisconsin) and Commonwealth v. Scantling (Massachusetts), where courts grappled with the admissibility and fairness of algorithmic risk assessment tools in criminal sentencing, directly confronting the bias and responsibility challenges inherent in AI adoption.⁶⁷

Effective AI governance necessitates the implementation of risk evaluation and compliance applications, enabling businesses to recognise, reduce, and manage risks posed by AI technologies while maintaining compliance with legal and ethical obligations. Risk assessment involves systematically evaluating possible hazards and weaknesses throughout the AI lifecycle—from data collection and model development to deployment and operation. This process locates sources of potential harm, including algorithmic bias, security flaws, data privacy violations, and unintended consequences, and evaluates their likelihood and potential impacts on relevant parties.

Another concern is the potential for AI systems to cause harm to people or society.⁶⁸ For instance, AI-powered medical diagnosis systems may generate false diagnoses, which

could potentially harm patients. Similarly, autonomous weapons systems by AI have the potential to make fatal judgments without human intervention, raising questions regarding ethics and morality.⁶⁹

The intersection of human rights and AI regulation has become increasingly critical, as AI technologies pose significant challenges to fundamental human rights like privacy, freedom of speech, equality before the law, and access to justice. Ensuring that AI development and deployment conform to the basic principles of human rights, particularly as AI becomes more integral to the criminal justice system, healthcare system, workplaces, and social services.

Concerns have arisen regarding the potential for AI systems to reinforce or amplify biases and inequities at the junction of human rights. The tension between AI deployment and fundamental rights is starkly illustrated in pending cases before international bodies. For instance, in AlgorithmWatch Schweiz and others v. Switzerland before the European Court of Human Rights, the plaintiffs challenge the lack of human review and transparency in fully automated systems used for significant public decisions. This case tests the boundaries of Article 8 (Privacy) and Article 6 (Fair Trial) of the European Convention on Human Rights.⁷⁰ AI algorithms trained on biased data may generate discriminatory outcomes, potentially resulting in unfair treatment and infringements on people's rights to equal protection under the law.

4.2.3. Economic Repercussions

Technology based on AI can drastically change a number of industries and open up new business opportunities,⁷¹ but it also prompts questions regarding the economic effects of its creation and application, notably with regard to the loss of jobs and the concentration of wealth.

AI can automate many operations currently performed by humans, leading to significant job displacement,⁷² especially in industries that involve repetitive tasks, such as manufacturing or data entry. Businesses that successfully adapt and use AI technology may experience an increase in productivity and profitability, potentially increasing wealth and market domination.⁷³ However, this may exacerbate economic inequality, concentrating wealth among a small number of businesses and individuals.

The economic impact of AI extends beyond particular businesses or industries. Widespread use of AI may have macroeconomic implications, including shifts in the labour market, changes in supply and demand, and variations in rates of economic expansion.⁷⁴

AI also challenges the traditional legal framework by generating new kinds of data and information.⁷⁵ Existing statutory laws often struggle to anticipate emerging issues in areas such as intellectual property, privacy, and data protection, which may therefore be overlooked and inadequately addressed.⁷⁶

Furthermore, AI has the potential to cast doubt on accountability under the law.⁷⁷ For example, it might not be apparent who is legally accountable for the harm caused if a medical diagnosis system powered by artificial intelligence makes a wrong diagnosis,⁷⁸ creating ambiguity and misunderstanding that statutory law may not be equipped to resolve.⁷⁹ High-profile cases, such as Waymo LLC v. Uber Technologies, Inc., though settled, underscore the complex liability and intellectual property challenges arising from rapid AI development, particularly in self-driving car systems, and highlight the difficulty statutory frameworks face in definitively assigning responsibility.⁸⁰

The fast pace and increasing complexity of AI development make it challenging for policymakers and legal professionals to enact laws capable of effectively governing AI creation and deployment.⁸¹

5 LIMITATIONS OF LEGAL POLICIES IN REGULATING DEVELOPMENTS OF AI

Two considerations must be mentioned at the outset. First, there are different degrees of conformity to the law. It is difficult to envisage a legal system composed solely of precise, mechanical principles—and it would likely be undesirable even if possible. Some degree of discretion will always remain in the legal system.

Second, as AI is increasingly used to support or replace human decision-making, the question arises as to what kind of process should be afforded to individuals affected by such judgments. The growing prevalence of black-box machine-learning algorithms renders many machine decision-making virtually unintelligible. This opacity is compounded by the phenomenon of "automation bias," whereby individuals exhibit overconfidence in the judgments made by machines and display prejudice against challenges to those determinations. Although AI is often promoted for its potential to reduce costs and increase efficiency, these benefits remain uncertain in the face of obstacles, particularly if they include significant procedural rights, such as transparency and due process, that are poorly safeguarded.

Although there have been attempts to manage and regulate the development of AI, there are still some legal constraints that need to be taken into consideration. These restrictions include, among others:

1. Inability to keep pace with technological change: AI technology evolves rapidly, whereas lawmaking is inherently slow. New laws take time to draft and put into effect, and existing laws can swiftly become obsolete.
2. Lack of global cooperation: The growth and implementation of AI transcend national boundaries. In the absence of international coordination, fragmented legal frameworks risk creating inconsistencies, enforcement gaps, and jurisdictional challenges. These difficulties are exacerbated when infringing parties are positioned in another country or when the legal system is precarious.
3. Ethical issues: The possibility of bias, discrimination, and privacy invasion are a few of the serious ethical issues raised by AI.⁸² These challenging ethical issues demand constant attention and assessment, and legal systems and regulations may find it difficult to handle them.
4. Limited knowledge of AI: Many legal professionals lack the technical know-how necessary to comprehend AI and its implications.⁸³ Their capacity to create and put into practice efficient legal regulations and policies may be constrained as a result.

6 INTERNATIONAL LEGAL DIFFICULTIES ARISING FROM THE USE OF AI

AI adoption generates a range of cross-border legal challenges that require careful attention.⁸⁴ In the United States, regulatory efforts regarding system-wide risk reduction in algorithmic decision-making have largely overlooked individual due process. There has been some agreement among recent legislative proposals from the United States; however, they have emphasised the need for systemic solutions rather than individual rights to contest, such as algorithmic impact studies or audits.

In contrast, regulators in Europe are approaching algorithmic decision-making from a comprehensive standpoint. The European Union's General Data Protection Regulation (GDPR),⁸⁵ which came into force in May 2018, provides certain individual rights for data subjects as well as systemic governance measures. Individuals whose decisions are subject to automated decision-making also have the ability to argue against specific findings. These rights also include access, transparency, amongst others. Similarly, the Council of Europe's updated data protection treaty specifies the right to appeal. The Council of Europe is an international organisation devoted to human rights, consisting of the member states of the European Union and a few non-EU nations.

Beyond Europe, the right to challenge AI outcomes is also gaining ground. The Organisation for Economic Co-operation and Development (OECD), an intergovernmental body influential in shaping global data protection regulations through its recommendations, is expected to extend its influence into AI governance as well. The "right to request a review of decisions taken" by AI is included in Brazil's extensive data protection law, enacted in 2018. Similarly, in November 2020, the Canadian Office of the Privacy Commissioner recommended amending Canada’s data privacy law to introduce a right to challenge AI conclusions.

Despite these developments, several significant legal challenges arise from the use of AI:

1. Intellectual property: AI systems may be protected under trade secrets, copyrights, or other types of intellectual property protection.⁸⁶ When using another person's intellectual property, developers must take care to respect their rights and secure the appropriate licenses or permissions.
2. Liability: AI raises complex questions of liability, especially in cases of accidents or errors brought on by algorithmic flaws, biases, or mismanagement. Developers and users must determine who should be held liable for any damage the systems may cause and ensure that sufficient insurance is in place to cover any potential risks. Traditional legal frameworks often struggle to sufficiently handle these unexpected difficulties, prompting calls for judicial precedents and clearer accountability standards.
3. Jurisdiction: As AI systems frequently operate across borders, they can lead to complicated jurisdictional difficulties. Developers and users must make sure they abide by the rules and laws of all pertinent jurisdictions.
4. Prejudice: AI systems trained on biased datasets risk generating discriminatory outcomes.
5. Trade restrictions: Where AI technologies have potential military or national security uses,⁸⁷ they may be subject to export controls and trade restrictions. Developers and users must therefore comply with all applicable export regulations. 7 CONCLUSIONS AND SUGGESTIONS

7 CONCLUSIONS AND SUGGESTIONS

The role of statutory law in regulating AI is fundamental in striking a delicate balance between fostering innovation and ensuring responsibility. As AI continues to evolve at a rapid pace, the establishment of robust legal frameworks becomes essential to address the ethical, social, and economic implications that accompany its development. Moreover, effective regulation must be dynamic and adaptive, reflecting the evolving nature of technology while maintaining core principles of transparency, privacy, and security.

Notice and transparency obligations under the General Data Protection Regulation (GDPR) for AI have garnered more attention, especially the so-called "right to explanation," which has ignited an upsurge of scholarly discourse. Though the GDPR explicitly establishes the right to dispute, regulators have not yet offered substantial guidance on the nature of the right or how it should be exercised.

The essence of democracy is under threat from the growing use of AI in decision-making. It is crucial to use design approaches that incorporate judicial review concepts as a fundamental component of AI-driven architecture in order to restore human confidence in AI. However, AI cannot wholly replace human bias and is therefore not always accurate; instead, it may obscure bias behind layers of purportedly impartial mathematical authority. Algorithmic outcomes can be biased even when programmers do not intend to discriminate, and these problems manifest across diverse technologies. For example,

actuarial algorithms used in criminal sentencing— despite their simplicity—have been shown to perpetuate bias, discrimination, and inaccuracy.

AI decision-making further raises questions about "what it means to be human." By excluding human judgment, empathy, and contextual reasoning, both public and private institutions risk reducing individuals to numerical values. It is arguable that the dignity of the human subject of the judgment is compromised when human decision-makers are replaced by machines. Yet, it would be a mistake to assume that algorithms are inherently more flawed than human decision makers; judges, too, may act with prejudice or inaccuracy. The dignity of a human subject can also be harmed by discrimination by a human decision-maker.

However, the transition from human to AI or hybrid human-AI decision-making systems fundamentally changes the policy environment and its underlying values. For instance, AI decision-making transfers some policy decisions early on to algorithm designers, rather than allowing a human decision-maker to assess a specific individual's unique circumstances ex post. In many cases, policy choices remain embedded in the "black box" of the algorithm, potentially opaque even to its creators. Decision-making processes and who decides what vary. The lack of transparency undermines accountability and the outcomes of the decisions.

The transition from individual customisation to decisions based on categories may also accompany a shift in AI decision-making. This gives rise to an issue known as the "long-tail problem," when an AI incorrectly classifies "weird stuff that is hard to deal with" into familiar categories. For instance, a self-driving car that has been taught to stay away from deer, cats, and dogs might not be able to "see" kangaroos crossing the road. A fraud warning algorithm used by the United States Department of Agriculture for the Supplemental Nutrition Assistance Program was trained to detect fraudulent activity on whole-number purchases; however, it failed to detect fraudulent activity at Somali-American grocers, where clients would buy meat in whole dollars. In actuality, the "long tail" can contain items that are not objectively considered "weird": Inappropriate consideration of illnesses like cerebral palsy or diabetes, which are hardly anomalies, was made by the erroneous home health care allocation algorithms.

A multidimensional and cooperative effort among several stakeholders—including developers, regulators, legislators, and civil society organisations— is necessary for a comprehensive strategy to govern AI that accounts for its ethical, social, and economic consequences. Such a strategy should include the following components:

• Fairness, accountability, transparency, and responsibility should be given top priority when developing and deploying AI systems. Their design and implementation should respect human rights, embrace diversity, and promote the welfare of society, as recommended by programmers and regulators.
• Public engagement is essential to ensure that the needs and issues of all stakeholders are addressed. This may involve consultation with community organisations, civil society organisations, and those who may be impacted by the use of AI systems.
• Data governance should be regulated by specific regulations on collection and use, supported by authorisation frameworks that safeguard privacy. Data protection and privacy regulations must be followed by those who build and deploy AI systems.
• Human oversight is critical to guarantee that they are operated safely and responsibly. This may entail deploying "human-in-the-loop" technologies, which permit people and AI systems to collaborate on decision-making.
• Risk assessment and management should be integral when developing and deploying AI systems. Developers and regulators must assess the ethical, societal, and economic implications of AI systems and implement mitigation plans accordingly.
• Interdisciplinary collaboration among experts from a range of fields, including computer science, law, ethics, social sciences, and the humanities, ensures that broader societal impacts of the employment of AI systems are considered.
• International coordination and cooperation are necessary to address the global nature of AI. In order to advance ethical and responsible AI, governments and civil society organisations should work together to establish international standards and recommendations that promote ethical and responsible AI.
• Regulatory reform is required, such as amending existing AI regulations (e.g., EU AI Act) to mandate AIAs for all high-risk public-sector AI and private systems in healthcare, hiring, finance, and criminal justice.

By embracing this balanced approach, statutory law can play a crucial role in guiding AI towards a future where technological advancement and societal well-being are not mutually exclusive, but are instead harmoniously integrated. In doing so, we can harness the full potential of AI to drive progress and innovation while upholding our collective responsibility to ethical standards and human values. Ultimately, AI systems should be developed and implemented in ways that benefit all members of society and promote responsible, transparent, and ethical algorithms in line with principles of fairness, public participation, data governance, risk assessment, and international collaboration.

FOOTNOTES

REFERENCES

AUTHORS INFORMATION

RIGHTS AND PERMISSIONS

EDITORS

ABOUT THIS ARTICLE

ADDITIONAL INFORMATION

DETAILS FOR PUBLICATION

AI DISCLOSURE STATEMENT

АНОТАЦІЯ