UN Creates Scientific Panel on AI Governance
The United Nations has created the Independent International Scientific Panel on Artificial Intelligence to advise on global governance. The move comes as AI systems increasingly surpass human-level benchmarks in complex tasks, according to the Stanford 2025 Index. The panel will analyze AI's societal impact, monitor risks, and recommend safeguards to balance innovation and ethics.
This panel follows the recommendations of the UN's High-level Advisory Body on AI, whose 2024 report "Governing AI for Humanity" identified a "global governance deficit" and a fragmented patchwork of rules. The report highlighted that 118 countries were not part of any significant international AI governance initiatives, creating a risk of disconnected regulatory systems. The 40-member panel is composed of experts from academia, the private sector, and civil society across all five UN regions. Notable members include Nobel Peace Prize laureate Maria Ressa, alongside specialists in data science, high-energy physics, and linguistics, selected from over 2,600 candidates. Its primary task is to produce an annual, evidence-based assessment of AI's risks and opportunities, synthesizing existing research. This work is designed to be policy-relevant but non-prescriptive, creating a common factual grounding for international discussions. The urgency is driven by rapid advancements, with AI performance on new, complex benchmarks like GPQA and SWE-bench improving by as much as 67 percentage points in a single year. At the same time, the cost to achieve high-level performance has plummeted, making advanced AI more accessible than ever. Key risks under consideration include algorithmic bias leading to discrimination in hiring or law enforcement, the mass proliferation of misinformation through deepfakes, and large-scale job displacement due to automation. The panel aims to provide a scientific consensus on the severity and trajectory of these threats. This UN initiative joins a landscape of other international efforts, including the G7's Hiroshima Process, the OECD's AI principles, and a series of AI Safety Summits that produced the Bletchley and Seoul Declarations. The panel's work is intended to complement, not replace, these existing forums. The panel's annual report will directly inform a new, parallel body: the Global Dialogue on AI Governance. This dialogue will serve as a forum for all 193 UN member states to share best practices and coordinate governance approaches based on the scientific panel's findings.
