22 countries back UN declaration for human control over AI

22 countries adopted a declaration Monday at the UN General Assembly stating that artificial intelligence must remain under human direction, oversight, and control, while proposing a global supervisory regime. The declaration acknowledges AI's benefits but warns that rapid development of frontier models poses serious safety and security risks. It urges mandatory pre-deployment testing and independent evaluation, though it stops short of calling for a slowdown or pause in AI development. Finnish President Alexander Stubb, a main organizer, hopes to gain U.S. and Chinese support, noting both leaders meet Thursday. He prefers creating an IAEA-like institution within the UN framework, while Norway's Prime Minister suggested the G7 and G20 could advance the initiative.