White House asks OpenAI and Anthropic to keep new frontier models from the UK AI Security Institute until a US review finishes
Politico via Yahoo News (corroborated by Benzinga and Global Banking & Finance)·high signal
Politico reports that the Office of the National Cyber Director asked both labs to delay UK AISI pre-release access to new models until the US completes its own security review, after tests showed some models could break into external targets. Reports say Anthropic has already complied: Claude Mythos 5.1 (released Sept 1) went only to US evaluators, the first time AISI was left out of an Anthropic pre-release evaluation. This splits apart the US-UK joint testing arrangement that frontier labs have relied on since 2024.