5th NIC-NIDA Conference, 2026

The rapid diffusion of artificial intelligence since the mid-2010s, and the public arrival of large generative models from 2022 onward, has outpaced the regulatory capacity of most states and international institutions. Early governance efforts were largely voluntary and sectoral: the OECD AI Principles (2019), the G20 AI Principles, UNESCO's 2021 Recommendation on the Ethics of AI, and national strategies emphasizing innovation. As capabilities scaled, concerns broadened from narrow questions of bias and privacy to systemic risks—labor displacement, election integrity, autonomous weapons, concentration of compute, and the erosion of state sovereignty over information ecosystems. Since 2023, governance has moved toward harder instruments and competing models. The European Union adopted the AI Act in 2024, creating a tiered, risk-based regime with extraterritorial reach. The United States issued Executive Order 14110 in 2023 (substantially revised in 2025), relying on agency action and voluntary commitments from frontier developers. China has implemented binding rules on recommendation algorithms, deep synthesis, and generative AI, while pursuing AI as a pillar of industrial policy. At the multilateral level, the UN Secretary-General's High-Level Advisory Body on AI delivered its final report in September 2024, and the Global Digital Compact, adopted at the Summit of the Future, committed states to an International Scientific Panel on AI and a Global Dialogue on AI Governance. The UK-initiated AI Safety Summit series (Bletchley 2023, Seoul 2024, Paris 2025) produced shared declarations but no enforcement architecture. The core disputes today concern who governs, what is governed, and to what end. States diverge on whether to prioritize safety of frontier models, equitable access for the Global South, protection of fundamental rights, or national competitiveness. Developing economies argue that current fora are dominated by a small set of compute-rich states and firms, while smaller advanced economies seek interoperability between the EU, US, and Chinese regimes. Unresolved questions include liability for AI-generated harms, governance of dual-use foundation models, compute and data sovereignty, AI in military decision-making, and the institutional home of any future global oversight body.