UN Launches Global AI Safety Panel
The United Nations has launched the Independent International Scientific Panel on Artificial Intelligence to provide governments with science-based guidance on AI's risks and opportunities. The panel is modeled on the Intergovernmental Panel on Climate Change (IPCC) and will evaluate safety, ethics, and societal impacts to help shape global norms and regulation.
The 40 members of the new AI panel were selected from a pool of over 2,600 candidates and will serve three-year terms in their personal capacities, not as representatives of their institutions or governments. The group is geographically and gender diverse, comprising experts from academia, the private sector, and civil society with backgrounds in machine learning, data governance, cybersecurity, and human rights. The panel's creation was approved by the UN General Assembly with a vote of 117 in favor and two against, with two abstentions. The United States and Paraguay voted against the measure, with a US representative calling the panel an "overreach of the U.N.'s mandate and competence" and expressing concern that it could be influenced by "authoritarian regimes." Ukraine and Tunisia abstained from the vote. Ukraine's abstention was an objection to the inclusion of a Russian expert on the panel. Despite the opposition, many US allies in Europe and Asia voted in favor of establishing the panel. Like the IPCC, the AI panel will not conduct its own research but will synthesize existing scientific findings to produce comprehensive assessments. This model is designed to create a shared, evidence-based understanding to inform international policymaking, though the panel itself has no power to create binding rules or regulations. The panel's first report is expected in July 2026. Its findings will be a key input for the broader Global Dialogue on AI Governance, a separate UN initiative aimed at fostering cooperation and shared frameworks among member states. This UN initiative joins a landscape of other international efforts to govern AI, including the G7's Hiroshima Process, the Global Partnership on AI (GPAI), and the Council of Europe's legally binding treaty on AI. The UN's approach is intended to be more globally inclusive, as 118 countries are not part of any other major international AI governance initiative.
