Alliance University

Understand the realities of Artificial Intelligence (AI) and how it has begun to question some of the basic structures of law itself particularly the issue of legal personhood and liability. When AI systems and services move across jurisdictions, what should be the baseline principles governing them? These must be developed in a way that does not limit technological progress, yet also does not require a complete reworking of the foundational structures of law. There is also a need to examine the emergence of binding AI frameworks with extraterritorial effects. When one State's regulatory model extends beyond its territory without the consent of other States, it raises fundamental questions of sovereignty that must be clearly addressed. At the same time, there is a need to consider the possibility of a common international framework or treaty that can bring a degree of coherence, without imposing uniformity. Universal educational and training programmes on AI should be prioritised, alongside serious and structured discussions on its use in sensitive sectors such as medicine and the military. Finally, AI governance and AI ethics guidelines must continue to evolve in a way that is both practical and globally relevant, ensuring that innovation and responsibility develop together.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

In March 2026, we have already seen how fast AI can merge into everyday life. From children using it as a therapist to taking extreme, life-threatening steps. Job markets hiring and restructuring. Open-source systems moving towards integration with a State's defence ministry. All of this shows a pattern of how a simple curiosity can turn into a mind-influencing force, leading basic principles of the UN Charter, human rights, and ethics to be rethought around this new-age phenomenon which some assume could lead to human extinction, while others see it as a futuristic step towards evolution.

Beyond these already critical concerns, there is a deeper layer that is still not being discussed enough is the shift from AI as a tool to AI as an environment shaping human behaviour. AI literacy, defence use, privacy risks, misinformation, deepfakes, and harmful applications such as non-consensual content are all visible manifestations. Beneath them lies a more structural issue. AI is no longer only executing tasks, it is increasingly influencing how humans think, decide, and perceive reality. This introduces a sharper concern. Not just misuse of AI, but dependence on AI-mediated reasoning. When individuals begin to rely on AI systems for emotional support, decision-making, or interpretation of information, the question is no longer about accuracy alone. It becomes a question of cognitive autonomy. Existing legal frameworks whether in privacy, human rights, or international law are not fully equipped to address influence without coercion. There is no clear doctrine for when assistance becomes dependence, when suggestion becomes manipulation, or when optimisation begins to shape human agency. This becomes particularly significant in high-stakes environments such as conflict, where AI may shape strategic narratives; in media, where perception can be engineered at scale; and in everyday life, where trust in systems may quietly replace independent judgment. The sharper edge, therefore, AI governance is not only about controlling systems. It is about preserving human agency within systems designed to optimise behaviour. This is a layer that sits beneath all others and one that law has only just begun to recognise.

In India, the most visible challenge is the gap between rapid adoption and evolving governance capacity. AI is already being integrated across sectors such as education, public services, finance, and media, often faster than formal regulatory frameworks can keep pace. Current developments including discussions around a Private Member's Bill in the Rajya Sabha (March 2026) on AI use in medical and research contexts indicate that the approach remains largely guideline-driven and sector-specific, rather than a comprehensive regulatory framework. This creates uncertainty around accountability, liability, and enforcement, particularly in high-risk areas like healthcare. At the same time, this presents an opportunity. India is positioned to develop flexible, adaptive governance models that can respond to scale and diversity without prematurely restricting innovation. The focus is gradually shifting toward building practical standards rather than imposing rigid ex-ante controls. In contrast, Europe reflects a more structured regulatory maturity, particularly through the EU AI Act. The emphasis is on risk classification, compliance obligations, and fundamental rights protection, providing a clearer legal framework for AI deployment. However, this also introduces challenges. The compliance burden may impact innovation, and the extraterritorial reach of EU regulation raises questions for non-EU actors, including Indian entities operating within European markets. Taken together, these developments highlight a broader pattern. India is moving through incremental, sector-led governance, while Europe is advancing through comprehensive regulatory architecture. The key challenge is managing this divergence without deepening fragmentation. The opportunity lies in creating interoperable approaches that allow flexibility in development while maintaining shared baseline standards across jurisdictions.

At present, international cooperation is limited less by disagreement on principles and more by divergence in how those principles are operationalised. A meaningful dialogue can act as a space to translate between regulatory architectures for example, aligning the EU's compliance-driven model, the United States of America innovation-led approach, and emerging economies' adaptive frameworks. Beyond this, the Dialogue can address a deeper and less discussed gap on the absence of a shared understanding of human agency in AI-mediated environments. Cooperation cannot be limited to technical standards alone; it must also engage with how AI systems influence decision-making, perception, and autonomy across societies. A further role lies in developing interoperable governance tools not binding rules, but shared templates for risk assessment, documentation, and oversight. This would allow systems to move across jurisdictions without requiring full regulatory harmonisation. Importantly, the Dialogue should also function as a mechanism for capacity equalisation, ensuring that developing states are not only rule-takers but active participants in shaping governance norms. Ultimately, its value lies in shifting cooperation from abstract alignment to practical coordination, while preserving legal diversity. The sharper edge is this: cooperation in AI governance is no longer about agreeing on rules, but about agreeing on how differences are managed within a shared system.

The AI Dialogue should build upon existing frameworks such as the OECD AI Principles (2019), UNESCO Recommendation on the Ethics of AI (2021), and the emerging regulatory architecture of the EU AI Act (2024). It should also connect with sectoral and operational initiatives, including due diligence models, risk management frameworks, and cross-border data governance discussions. However, the added value of the AI Dialogue lies not in replicating these efforts, but in addressing what they do not fully capture. Most existing initiatives operate at the level of principles or regulation, but there is a missing layer: how governance functions in practice across fragmented systems. The Dialogue can bridge this by focusing on interoperability between frameworks, rather than creating new ones. A sharper and less explored contribution would be its ability to engage with grey zones areas where law, policy, and technology overlap but do not fully align. This includes issues such as AI influence on cognition, responsibility in human-AI collaboration, and the use of AI in conflict and strategic environments. Additionally, the Dialogue can act as a platform to integrate non-state actors including private developers, researchers, and civil society into governance conversations without displacing state responsibility. Its real added value, therefore, is not standard-setting alone, but system-connecting: linking existing initiatives into a more coherent, multi-layered governance ecosystem. The sharper edge is this: the future of AI governance will not be defined by new frameworks, but by how effectively we connect the ones that already exist.

Different stakeholders should not only participate, but contribute through clearly differentiated roles aligned with their capacities. States remain central in defining legal frameworks and ensuring accountability. However, private actors particularly developers and platform providers contribute through technical design choices, deployment practices, and operational governance systems. Academia and research institutions can anchor the Dialogue in critical analysis, evidence-building, and long-term thinking, while civil society brings in perspectives on rights, harms, and lived realities. The structure of the Dialogue should reflect this plurality. Rather than a single forum, it should operate as a multi-layered process a high-level intergovernmental track for normative alignment. Technical working groups for operational tools and standards. Cross-sector forums to translate policy into practice. A key recommendation is to move away from purely declaratory formats toward problem-oriented engagement where stakeholders jointly examine specific use cases (e.g., AI in healthcare, defence, or media) and identify governance responses. The sharper edge is this participation should not be symbolic. Each stakeholder should be linked to specific governance functions, ensuring that dialogue translates into responsibility and action.

Global AI governance discussions continue to underrepresent developing countries, the Global South, and smaller regulatory jurisdictions, particularly those most affected by imported AI systems but least involved in their design. Beyond geography, there is also underrepresentation of practitioners implementing AI governance on the ground. Communities directly impacted by AI harms, Interdisciplinary voices bridging law, technology, and society A less visible gap is the absence of perspectives on human cognition and behavioural impact how AI systems influence decision-making, perception, and autonomy in everyday life. Inclusion requires more than invitations. It requires structural integration of dedicated participation channels for underrepresented regions, capacity-building support to enable meaningful engagement and incorporation of local case studies and lived experiences into discussions. The Dialogue should also ensure that participation is not limited to those already aligned with dominant regulatory models.

One approach is the use of real incidents, where stakeholders respond to AI deployment in conflict, healthcare decision-making, or misinformation crises. This allows governance challenges to be explored in context, rather than in abstraction. Another format is "decision-chain mapping", where participants trace how an AI system moves from development to deployment to impact. This helps identify where responsibility lies and where governance mechanisms fail. Interactive cross-jurisdictional labs could also be introduced, where different regulatory models are applied to the same case to test interoperability and identify friction points. Importantly, formats should allow for controlled disagreement, rather than forced consensus. Structured debates and adversarial panels can surface tensions that are often overlooked in diplomatic settings. The deeper layer is meaningful engagement is not created through more discussion, but through exposure to complexity and constraint. The sharper edge is that AI governance cannot be fully understood in theory. It must be experienced through structured interaction with real-world dilemmas.

At the regulatory level, the EU AI Act (2024-2025) offers a structured, risk-based approach, translating principles into enforceable obligations such as documentation, conformity assessments, and post-market monitoring. In contrast, the United States continues to adopt a more market-led, sectoral approach, relying on executive guidance, agency-level rules, and industry standards. China reflects a state-centric model, with targeted regulations on algorithms and generative AI, emphasising control, security, and content governance. Other jurisdictions illustrate further variation. The United Kingdom has taken a pro-innovation, regulator-led approach, avoiding a single comprehensive statute in favour of existing sectoral regulators. Australia is moving toward risk-based and safety-focused frameworks, while South Korea combines strong digital infrastructure with emerging AI governance standards. Taiwan is positioning itself through technology-driven development with democratic oversight, particularly in data governance and semiconductor-linked AI ecosystems. Alongside these regulatory models, the OECD AI Principles (2019) and the UNESCO Recommendation on the Ethics of AI (2021) provide widely accepted normative baselines, creating a shared language across these diverse systems. At the operational level, AI due diligence frameworks based on risk identification, mitigation, auditability, and traceability are becoming central. These embed governance within organisational processes. Similarly, human-in-the-loop and human-in-command models are being adopted in high-risk sectors to retain accountability. A key emerging approach is interoperability-focused governance, enabling systems to function across jurisdictions despite regulatory fragmentation. AI governance is not converging into a single global model. It is evolving as a plural, multi-system architecture. The future of AI governance will depend less on creating uniform rules, and more on managing the friction between fundamentally different governance philosophies.