US Rejects Global AI Governance; 20+ Leaders Demand Mandatory Standards

On September 23, 2026, White House science adviser Michael Kratsios told the UN Security Council that the US "totally rejects any attempt to construct a globalist scheme of control for superintelligence," urging nations to build their own technical capacity and pursue bilateral agreements instead. Around the same time, an open letter signed by more than twenty heads of state, including the prime ministers of Australia, Canada, Germany, Finland, Norway, Singapore, and South Africa, called for mandatory pre-deployment testing with independent evaluators, common international safety standards, and a new UN-affiliated institution capable of setting standards and verifying compliance when capability thresholds are crossed.

The US position has been articulated consistently across multiple international forums. At the India AI Impact Summit in February 2026, Kratsios promoted "AI sovereignty" as the alternative to multilateral governance, describing it as nations controlling their own AI infrastructure, data, models, and policies. He argued that safety-focused regulation "entrenches incumbents and isolates developing countries from full participation in the AI economy" and warned that invoking safety can enable "tyrannical control." The American AI Exports Program is described as the vehicle for bilateral AI diplomacy. In October 2025, Kratsios described the US-Japan Technology Prosperity Deal as a model for this bilateral approach, calling international diplomacy the third pillar of the US AI Action Plan.

French President Macron, speaking at the India summit, offered a contrasting position, expressing determination to ensure safe oversight of AI and to position Europe as a safe space for AI innovation. The open letter from the twenty-plus heads of state cites "capable AI systems circumventing testing safeguards, exploiting vulnerabilities and gaining unauthorized access to real-world systems" as the basis for urgency.