Mustafa Suleyman, head of Microsoft AI, stated that competition with China should not delay the implementation of artificial intelligence safety measures. In a CNN interview published on September 20, he urged the United States to develop independent control mechanisms for increasingly powerful AI systems while maintaining cooperation with China where interests align.
Suleyman advocated for systems where humans retain the ability to monitor AI operations, interrupt processes, and verify decisions. He also supported independent model evaluations and international safety agreements. These remarks represent his public stance rather than a new corporate policy or a formal legislative proposal.
Regarding independent verification, Suleyman noted that external evaluators require sufficient access to systems to inspect behavior, restriction mechanisms, and internal controls. This process necessitates defining who conducts audits, what data is shared, and how companies must respond to identified issues.
The position emerges amidst US debates over whether strict safety requirements might hinder industry growth relative to China. Suleyman suggests decoupling these issues, arguing that technical stop mechanisms, secure audit logs, and independent verification can progress regardless of the speed of international negotiations.
