Bal Legal, Uskudar University

The success of the first Global Dialogue on AI Governance will depend on its ability to move beyond general principles and address concrete legal uncertainties in high-risk domains, particularly within criminal justice systems. First, the Dialogue should develop clear and operational accountability frameworks for AI-assisted decision-making. In contexts where AI directly affects individual liberty, the question of "who is responsible?" is no longer theoretical but urgently practical. Existing legal systems struggle to determine causation and fault within distributed, multi-actor algorithmic processes. Therefore, the Dialogue must advance models in which responsibility is not diffused, but clearly attributable and traceable. Second, the issues of algorithmic transparency and evidentiary validity must be addressed. For AI outputs to be admissible and reliable in criminal proceedings, technical accuracy alone is insufficient. These systems must also be legally contestable, auditable, and open to challenge. Otherwise, the use of opaque "black box" systems risks systematically undermining the right to a fair trial. Third, the Dialogue should recognize that AI is not merely a tool, but in certain contexts may become a subject of legal responsibility, raising fundamental questions for criminal law and necessitating a paradigm shift in legal doctrine. Finally, to prevent deepening global inequalities, the Dialogue should promote integrated capacity-building mechanisms that strengthen forensic, technical, and legal infrastructures, enabling developing countries to act not only as users, but also as norm-setters in AI governance. Ultimately, true success will lie in establishing a governance framework that not only enables AI innovation, but also effectively constrains it in order to safeguard justice.

• Safe, secure and trustworthy AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

My selection reflects a focus on the legal and societal risks posed by the deployment of artificial intelligence in high-impact domains, particularly within criminal justice systems. First, ensuring safe, secure, and trustworthy AI is fundamental where algorithmic systems influence decisions affecting individual liberty. In such contexts, technical reliability alone is insufficient; AI systems must operate within robust legal and ethical boundaries to prevent harm and injustice. Second, the protection and promotion of human rights is a central priority. The use of AI in areas such as surveillance, risk assessment, and evidence evaluation raises serious concerns regarding due process, equality before the law, and the right to a fair trial. Without strong safeguards, AI systems risk reinforcing existing biases and producing discriminatory outcomes. Third, transparency, accountability, and human oversight are essential to ensure that AI-driven decisions remain contestable and subject to meaningful review. In criminal justice settings, opaque "black box" systems undermine the ability of individuals to challenge decisions, thereby weakening fundamental procedural guarantees. Clear accountability frameworks are also necessary to determine responsibility in cases of error or harm. These priorities are closely interconnected. Trustworthy AI cannot exist without human rights protections, and both depend on effective transparency and accountability mechanisms. Addressing these areas is therefore critical to developing governance frameworks that not only enable innovation, but also safeguard justice and uphold the rule of law.

Yes. In addition to the existing themes, the following emerging issues should be explicitly recognized as distinct thematic priorities: 1. Responsible AI (as a legally enforceable standard) Responsible AI should evolve beyond a general ethical concept into a legally operational framework. In high-stakes domains such as criminal justice, responsibility must be clearly defined, traceable, and enforceable. This includes the ability to identify decision-makers, allocate liability, and ensure that AI-assisted outcomes can be effectively challenged and reviewed. 2. Evidentiary Status of AI Outputs and Chain of Custody in AI Systems AI-generated and AI-processed data raise fundamental questions regarding admissibility, reliability, and integrity as legal evidence. There is a critical need to establish standards governing how such outputs are introduced, evaluated, and contested in judicial proceedings. In parallel, maintaining a secure and verifiable chain of custody for AI-related data is essential to prevent manipulation and to preserve evidentiary authenticity. 3. Cross-border Use and Jurisdictional Challenges in AI Systems The transnational nature of AI systems creates significant jurisdictional ambiguity. When systems operate across multiple legal regimes, determining applicable law, assigning responsibility, and enforcing decisions becomes increasingly complex. This necessitates the development of coordinated international frameworks and interoperable legal approaches. These themes are essential to ensure that AI governance frameworks are not only principled, but also legally robust, practically applicable, and capable of addressing the complexities of real-world deployment.

As someone living and working in both Türkiye and Qatar, I observe that governance gaps in safe, secure and trustworthy AI, protection of human rights, and transparency, accountability, and human oversight create both challenges and opportunities. In Türkiye, the main challenge lies in the lack of binding and operational standards to ensure trustworthy AI. While awareness is increasing, gaps in accountability and oversight frameworks create uncertainty, particularly in high-risk areas such as law enforcement and judicial processes, where fundamental rights may be affected. In Qatar, rapid digital transformation and strong investment in AI have accelerated adoption. However, this has also exposed gaps in transparency and accountability, especially in explaining and reviewing AI-assisted decisions. Ensuring effective human oversight remains a key concern. In both countries, insufficient transparency and weak accountability mechanisms may risk undermining human rights, including due process and fairness. At the same time, these challenges present opportunities. Both Türkiye and Qatar can develop forward-looking governance frameworks that integrate legal, technical, and ethical perspectives, strengthening trust and ensuring that AI systems align with fundamental rights. Ultimately, addressing these gaps is essential to ensure that AI remains both innovative and just.

The AI Dialogue can play a pivotal role in advancing international cooperation by serving as a platform for convergence between diverse legal, technical, and policy approaches to AI governance. First, it can facilitate the development of shared baseline principles and interoperable frameworks, helping to reduce fragmentation across jurisdictions. Given the transnational nature of AI systems, aligning standards on accountability, transparency, and human rights is essential to ensure consistency and legal certainty. Second, the Dialogue can promote mutual learning and capacity-building, particularly by bridging the gap between developed and developing countries. By sharing best practices, technical expertise, and regulatory experiences, it can support more inclusive participation in shaping global AI governance. Third, it can act as a space for multi-stakeholder engagement, bringing together governments, academia, the private sector, and civil society. Such collaboration is critical to ensuring that governance frameworks are not only technically sound, but also socially legitimate and rights-oriented. Additionally, the Dialogue can contribute to the establishment of coordinated approaches to cross-border challenges, including jurisdictional issues, data governance, and enforcement mechanisms. In the long term, the AI Dialogue can move beyond discussion and function as a catalyst for actionable cooperation, enabling the co-creation of governance models that are both globally relevant and adaptable to local contexts.

The AI Dialogue should build upon existing initiatives not only at a normative level, but by developing concrete and operational mechanisms for cooperation. First, joint working groups could be established across existing frameworks such as the OECD, UNESCO, and the EU to develop harmonized standards on transparency, accountability, and human rights. This would help reduce regulatory fragmentation and promote consistency across jurisdictions. Second, in high-risk domains such as criminal justice, joint pilot projects and case-based collaborations could be implemented to test how governance principles function in practice. This would enable a transition from abstract principles to evidence-based regulatory approaches. Third, considering the transnational nature of AI systems, mutual recognition frameworks could be developed to facilitate the cross-border acceptance of systems that meet agreed standards. In addition, to strengthen auditability, traceability, and intervention capacity in AI systems, the Dialogue should support the creation of expert working groups. These groups could develop both technical and legal tools to investigate algorithmic decisions, identify errors, and enable timely intervention where necessary. At the same time, the Dialogue should encourage the development and empowerment of civil society platforms. Such platforms can play a critical role in providing independent oversight, enhancing transparency, and strengthening public trust in AI systems. These efforts should be supported by capacity-building initiatives, including training programs, expert exchanges, and knowledge-sharing mechanisms. Through these mechanisms, the AI Dialogue can move beyond discussion and become a practical instrument for developing accountable, auditable, and trustworthy AI governance frameworks.

Different stakeholders can contribute to the AI Dialogue by bringing complementary expertise and perspectives that reflect the multi-dimensional nature of AI governance. Governments should provide regulatory direction, share national experiences, and support the development of interoperable legal frameworks. The private sector can contribute technical knowledge, implementation experience, and insights into system design and risk management. Academia plays a critical role in advancing independent research, developing conceptual models, and evaluating the societal and legal impacts of AI systems. Civil society organizations are essential for ensuring inclusivity, safeguarding human rights, and providing independent oversight. To ensure meaningful participation, the AI Dialogue should adopt a structured and multi-layered format. First, it should establish thematic working groups focused on key areas such as accountability, transparency, and high-risk applications. These groups should produce concrete outputs, including policy recommendations and model frameworks. Second, the Dialogue should include case-based sessions where stakeholders analyze real-world scenarios, particularly in sensitive domains such as criminal justice, to bridge the gap between theory and practice. Third, the process should incorporate expert panels and technical forums to address complex issues such as auditability, traceability, and cross-border governance. Additionally, multi-stakeholder roundtables should be organized to ensure balanced representation and inclusive dialogue. Finally, the Dialogue should be designed as a continuous and iterative process, supported by digital platforms for ongoing collaboration, written submissions, and knowledge sharing. Such a structure would enable the AI Dialogue to move beyond general discussions and produce actionable, inclusive, and practically relevant governance outcomes.

Several important voices and perspectives remain underrepresented in global discussions on AI governance, limiting the inclusiveness and effectiveness of emerging frameworks. First, practitioners in high-impact domains, particularly from criminal justice systems—such as judges, prosecutors, defense lawyers, and forensic sciences experts—are often excluded. Yet, these actors directly engage with the real-world consequences of AI-assisted decisions and can provide critical insights into accountability, evidentiary standards, and due process. Second, Global South countries and developing regions are frequently underrepresented. This risks reinforcing existing inequalities, as governance models may be shaped without adequately reflecting diverse socio-economic and legal contexts. Third, civil society organizations with technical and legal expertise remain insufficiently integrated into decision-making processes. Their role in independent oversight and rights-based advocacy is essential but often limited to consultative participation. Additionally, affected individuals and communities, particularly those subject to algorithmic decision-making (e.g., individuals involved in criminal justice processes), are rarely included in governance discussions. Their experiences are crucial for understanding the societal impact of AI systems. To address these gaps, the AI Dialogue should adopt inclusive participation mechanisms, including targeted invitations, regional consultations, and dedicated forums for underrepresented groups. It should also support capacity-building initiatives to enable meaningful engagement, particularly in developing countries. Moreover, establishing structured channels for civil society and practitioner input, as well as incorporating real-world case-based contributions, can ensure that governance frameworks are grounded in practical realities. A truly effective AI governance framework requires not only technical and policy expertise, but also the inclusion of those who are most affected by its outcomes.

To foster meaningful and dynamic engagement, the AI Dialogue should adopt interactive, practice-oriented, and multi-layered formats that move beyond traditional panel discussions. First, case-based simulation sessions can be highly effective, particularly in high-risk domains such as criminal justice. Participants could be presented with real or hypothetical AI-related scenarios and asked to collaboratively assess legal, ethical, and technical implications. This approach bridges theory and practice while encouraging active participation. Second, the Dialogue should include multi-stakeholder problem-solving labs, where small, diverse groups (including policymakers, technologists, legal experts, and civil society) work together to develop solutions to specific governance challenges. These labs can produce concrete outputs such as draft guidelines or policy recommendations. Third, interactive "audit and challenge" workshops could be introduced, allowing participants to examine AI systems or decision-making processes and identify potential risks, biases, or accountability gaps. This would strengthen practical understanding of transparency and oversight mechanisms. Fourth, the use of regional dialogue clusters can ensure that discussions reflect diverse legal, cultural, and socio-economic contexts, while still feeding into a global framework. Additionally, digital participation platforms should be integrated to allow continuous engagement beyond the main sessions, including written inputs, real-time feedback, and collaborative drafting. Finally, incorporating reverse panels, where affected communities and practitioners take the lead in discussions, can rebalance traditional power dynamics and enrich the Dialogue with grounded perspectives. Such innovative formats would transform the AI Dialogue from a passive exchange of views into an active, solution-oriented and inclusive governance process.

Beyond well-known regulatory frameworks, several emerging practices offer more innovative and operational approaches to effective AI governance. One important example is the development of algorithmic audit trails, which enable the reconstruction of how an AI system reached a specific decision. This approach goes beyond transparency by allowing systems to be forensically examined, particularly in high-stakes contexts such as criminal justice. Another promising practice is the use of AI impact litigation and legal stress-testing, where hypothetical or real cases are used to test how AI systems perform under legal scrutiny. This helps identify gaps in accountability, due process, and evidentiary standards before harm occurs. Regulatory co-design platforms also represent an innovative approach. These platforms bring together regulators, developers, and affected stakeholders to collaboratively design AI systems and governance rules from the outset, rather than regulating retrospectively. In addition, independent algorithmic oversight bodies-including hybrid public-civil society institutions-are emerging as mechanisms to ensure continuous monitoring and external accountability of AI systems. Another example is the use of "human override protocols", which define clear conditions under which automated decisions must be reviewed, suspended, or reversed by human authorities. Finally, cross-border audit cooperation mechanisms are increasingly relevant, enabling different jurisdictions to jointly assess and verify AI systems that operate transnationally. These approaches demonstrate that effective AI governance requires not only rules, but also practical, testable, and enforceable mechanisms that can adapt to the complexity of real-world AI systems.