regulatory drafting

AI safety vulnerabilities—including algorithmic bias and adversarial fragility—fuel misinformation, inequity, security threats, and eroded public trust, exposing critical gaps in existing governance frameworks. To address this, we propose a unified three-dimensional governance framework: “intrinsic safety,” “derivative safety,” and “socio-ethical alignment.” This is the first approach to holistically integrate technical defenses (e.g., robustness enhancement, fairness-aware modeling, adversarial detection), real-world risk assessment (leveraging emerging evaluation benchmarks), and cross-disciplinary policy coordination. Through a systematic review of over 300 studies, we identify three core challenges: generalization gaps, insufficient evaluation rigor, and regulatory fragmentation—arguing for proactive, lifecycle-integrated governance rather than reactive remediation. Our work delivers a tripartite output: actionable technical guidelines, standardized evaluation metrics, evidence-based policy recommendations, and an open-source tool suite to advance global trustworthy AI ecosystems.

This study addresses a critical gap in current AI governance, which predominantly emphasizes substantive rules while neglecting the legal and regulatory infrastructure necessary for their generation and implementation. For the first time, this work systematically positions legal infrastructure as the cornerstone of effective AI governance and proposes an institutional framework comprising a frontier model registration system, an autonomous agent identification mechanism, and a market-oriented regulatory service model. Through rigorous legal design, regulatory modeling, and policy mechanism analysis, the research delivers an actionable institutional pathway that significantly enhances the flexibility, scalability, and enforcement efficacy of AI governance rules.

Privacy laws designate “consent” as a lawful basis for data processing, yet its translation into software implementations has long suffered from a legal–technical gap and opaque development practices. This paper proposes the first LLM-based, three-step automated framework: (1) legal clause parsing, (2) use-case compliance classification, and (3) technical requirement reconstruction—augmented by human-in-the-loop verification to ensure legal alignment. It establishes the first systematic, end-to-end mapping from privacy regulation text to executable technical specifications, enabling compliance-aware requirements engineering and use-case remediation. Empirical evaluation demonstrates that the LLM effectively identifies and rectifies non-compliant use cases, validating its feasibility for automated compliance tasks; it also reveals persistent limitations in complex legal reasoning. The work introduces a novel paradigm and practical pathway for AI-augmented legal technologization.

This study examines how fundamental rights enshrined in the Charter of Fundamental Rights of the European Union can be effectively embedded and safeguarded within artificial intelligence governance. Through legal textual analysis, institutional framework assessment, and rights impact evaluation, the research demonstrates that the EU Artificial Intelligence Act’s risk-based regulatory framework treats fundamental rights not merely as normative values but as legally binding threshold conditions and procedural triggers throughout the entire AI system lifecycle. The work offers the first systematic account of the dual normative and procedural functions of fundamental rights in AI governance, proposing a “rights-centric” regulatory paradigm. It highlights the Act’s potential to serve as a global model for rights-protective AI regulation while identifying key challenges to its effective implementation.

Rapid AI advancement poses novel governance challenges, necessitating a rigorous, technically grounded approach to AI governance. Method: This work introduces “technical AI governance” as a distinct paradigm and establishes the first interdisciplinary analytical framework—integrating AI safety, mechanism design, policy modeling, and governance theory—to systematically address three core problem domains: risk identification, evaluation of intervention effectiveness, and compliance mechanism design. Adopting a problem-driven methodology, it clarifies how technical tools can concretely support governance practice. Contributions/Results: (1) A formal, structured definition of technical AI governance and a taxonomy of its core problems; (2) The first publicly available, extensible open-problems catalog for technical AI governance, bridging methodological gaps between technical and policy communities; and (3) An actionable, problem-oriented investment guide for researchers and funding agencies to prioritize high-impact technical governance research.

The European Union’s AI governance faces fragmentation across policy instruments, with tensions in scope, regulatory stringency, and priority setting. This study employs a mixed-methods approach—integrating qualitative thematic analysis with BERTopic-based unsupervised topic modeling—to systematically examine key post-2018 policy documents, including the AI Act and the High-Level Expert Group’s (HLEG) Ethics Guidelines. Methodologically, it advances interpretability by coupling explainable quantitative topic modeling with rigorous policy hermeneutics, expanding textual coverage and deepening semantic insight. Results reveal a clear diachronic shift from principle-based ethics toward risk-based regulation: foundational HLEG values—such as transparency and human oversight—are institutionalized in the AI Act, yet the emphasis pivots to enforceability and high-risk AI system oversight. The study thus offers both a methodological framework and empirical evidence for assessing policy coherence, diagnosing regulatory gaps, and strengthening transnational AI governance coordination.

This study addresses the persistent challenge of operationalizing AI governance requirements within software development practice, particularly at the team level. Through an embedded action research approach in an AI startup, the authors construct a translational pipeline that bridges regulatory texts and concrete engineering actions. They propose a governance implementation framework grounded in practitioners’ cognitive orientations—convergence, alignment with existing practices, and disengagement—to shift governance responsibility from externally imposed mandates toward collective team accountability. By integrating legal text analysis, cross-functional collaboration, and collective assessment, the project surfaces developers’ authentic attitudes toward regulation, identifies compliance priorities anchored in user and developer needs, and renders implicit governance work explicit and institutionalized.

This work addresses a longstanding gap in AI alignment research—the underutilization of law as a critical source of normative and technical constraints. It introduces a novel “legal alignment” paradigm that systematically integrates legal rules, interpretive methodologies, and institutional structures into AI development. The framework advances three core directions: formal modeling of legal norms, reasoning mechanisms grounded in legal interpretation, and compliance-oriented evaluation and governance architectures. By deeply embedding legal knowledge systems into the foundations of AI alignment, this study provides both theoretical grounding and technical pathways for building lawful, trustworthy, and collaboratively capable AI systems. Furthermore, it fosters interdisciplinary synergy between legal scholarship and artificial intelligence, enabling the institutional implementation of legal alignment in real-world applications.

Existing AI governance research predominantly emphasizes normative principles, lacking executable engineering mechanisms spanning the full AI lifecycle. This paper proposes a novel paradigm—“Responsibility as Closed-Loop Supervisory Control”—and introduces a six-layer control-theoretic architecture that formally encodes societal values—including fairness, autonomy, cognitive load, and explainability—into modelable, monitorable, and enforceable closed-loop constraints. Methodologically, it innovatively integrates safety envelope modeling, feedback-driven explanation frameworks, and end-to-end mapping of ethical objectives to control parameters, combining constrained optimization, runtime monitoring, behavioral interface design, and multi-tiered auditing. Empirical validation across clinical decision support, cooperative autonomous driving, and public-sector systems demonstrates real-time monitoring and dynamic assurance of normative objectives, enabling accountable, adaptive, and auditable AI deployment.

This paper addresses the challenge of clarifying and coordinating responsibilities among six key actor roles—providers, deployers, authorized representatives, importers, distributors, users, and notified bodies—under the EU Artificial Intelligence Act (Regulation (EU) 2024/1689). Applying systematic legal text analysis, the study examines all 113 articles, 180 recitals, and 13 annexes. It proposes a dynamic role-transformation mechanism grounded in the principle that “control determines responsibility,” revealing how obligations cascade along the AI supply chain. The paper’s primary contribution is an original AI regulatory actor mapping framework that operationalizes the “responsibility follows control” principle to enable distributed, cooperative governance. This framework balances fundamental rights protection with innovation support by clarifying role-specific duties and interdependencies. It delivers actionable guidance for regulators, AI providers, deployers, and other compliance stakeholders on role identification and obligation fulfillment under the AI Act.

This study addresses the legitimacy crisis arising from AI systems exercising governance authority without proper authorization. It introduces the sociological concept of legitimacy into AI governance, clearly distinguishing legitimacy from alignment and proposing three transferable principles—integrativeness, familiarity, and contestability—as its core. Drawing on legal-institutional analysis and sociological theory, the research demonstrates that neither technical performance nor formal compliance suffices to resolve AI’s authority deficit. Instead, it argues for constructing “thick legitimacy” through public authorization and procedural safeguards, thereby establishing legitimacy as an independent objective in AI governance.