The White House requested OpenAI and Anthropic refrain from sharing new models with the U.K. AI Safety Institute (AISI) until U.S. authorities complete initial reviews [7]. This directive aims to assert U.S. leadership in AI safety and ensure domestic priority access to cutting-edge American technology.
The U.S. government is prioritizing its confidential review framework over existing arrangements with allied nations [13]. This framework operates under a June 2026 executive order.
The policy shift occurs despite AI executives from these companies publicly advocating for increased international cooperation and global safety standards [16].