Twenty countries and the European Union called for advanced AI to remain under human oversight, comply with international law and face common standards and independent evaluation. Their proposal includes exploring an international institution that could set standards, enable verification and convene governments when capability thresholds are crossed. The United States and China did not join, leaving the two leading AI powers outside the initiative.
Twenty countries and the European Union issued an international call for advanced AI models to remain under human direction and oversight. Launched by Finnish President Alexander Stubb and Norwegian Prime Minister Jonas Gahr Støre, the initiative asks that these systems be developed safely and in line with international law. Germany, South Africa, Canada, Australia, the United Arab Emirates and Singapore are among its supporters.[1], [2]
The appeal proposes mandatory safety testing before companies deploy advanced models and enough access for independent evaluators to examine risks. It asks governments to share serious safety incidents, coordinate common standards and widen access to trusted evaluation capacity across regions. The proposed oversight would therefore include a common verification structure beyond companies' own accounts of their systems.[1], [2]
The signatories asked United Nations members to build on existing international mechanisms and explore creating a new institution. The proposed body could set standards, verify compliance and convene states when AI capabilities cross defined thresholds. The appeal remains open to further endorsements, while the United States and China are absent from the announced group of signatories.[1], [2]
