22 countries adopt declaration during UN General Assembly, calling for human control over AI and exploring international regulatory body

22 countries adopted a declaration during the UN General Assembly stating that AI must remain under human control, oversight, and governance, and recommending the establishment of global regulatory mechanisms. The declaration acknowledges AI's potential to improve lives, advance science, and strengthen economies, while warning that frontier AI models, if mismanaged, pose serious safety and security risks. The document requires AI companies to develop transparent safety protocols, including mandatory pre-deployment testing and independent evaluation, and to provide qualified assessors with sufficient access; it also calls on countries to coordinate common standards, share reports of serious safety incidents, and explore establishing an international body. The declaration does not call for slowing down or pausing AI development; organizers say the focus is on managing risks. The United States and China have not joined, and France, Italy, and Poland have not signed; the UK has not joined yet. Finnish President Stubb and Norwegian Prime Minister Støre are the main initiators; Stubb personally favors establishing an IAEA-like body within the UN framework but emphasizes keeping options open.