White House Tells OpenAI, Anthropic to Withhold New Models From UK Testing Body

The request puts frontier-model developers in the middle of a US–UK divergence over who validates advanced AI first, turning safety testing into a question of sovereign precedence rather than technical capacity. It affects both labs and the UK body, whose access to the newest systems now depends on Washington's undisclosed schedule. Whether allied testing arrangements can survive unilateral sequencing, and whether labs comply quietly or push back, is the open question.