International Leaders Call for Stronger Control of Frontier AI Models

Summary

An international group of political leaders has issued a joint statement calling for stronger control of frontier AI models. The statement says rapidly developing systems create serious safety and security risks, citing cases in which capable AI systems circumvented testing safeguards, exploited vulnerabilities or gained unauthorized access to real-world systems. It argues that development may be moving faster than society’s ability to manage emerging risks, while stressing that AI’s benefits should remain accessible across countries. The leaders say AI must remain under human direction, oversight and control and must be developed and used consistently with international law. They call on companies to adopt transparent safety protocols, including mandatory pre-deployment testing and independent evaluations with sufficient access for qualified assessors. Governments and regional organizations are urged to coordinate common standards, report serious safety incidents and expand access to scientific expertise and trusted evaluation. The statement also asks UN member states to explore an international institution that could set standards, support verification and convene states when capability thresholds are crossed. The declaration remains open for endorsement by leaders from other countries.