White House Asks OpenAI, Anthropic to Delay UK AISI Access

The Office of the National Cyber Director has asked OpenAI and Anthropic to keep new frontier models away from the UK's AI Security Institute until the US government finishes its own review, Politico reported. AISI is described in the reporting as one of the most well-resourced government testing agencies in the world, and one that previously enjoyed privileged early access.

Anthropic has already complied. Claude Mythos 5.1, released on September 1, 2026, went to a US-only slate under the company's Project Glasswing program, and it is the first time AISI has been left out of a pre-release Anthropic evaluation. The company said the model was "only available to a set of U.S. organizations" and that it would "expand access to a broader set of domestic and international partners as quickly as possible."

A senior administration official gave a short rationale: "Because they're American companies and this has been our policy with every new frontier model that comes out." The White House framing, per the reporting, is that it wants to make sure US systems are secure before models are shared with partners. The context includes a recent incident in which an OpenAI agent reached into an Australian government website during testing.

AISI Director Henry de Zoete acknowledged the gap while insisting the institute still holds "strong relationships with all frontier AI developers and continue to have prerelease access to some of the world's most capable models," naming OpenAI's GPT-6 Astra. A UK Cabinet Office spokesperson was blunter about the direction of travel: "These risks do not stop at national borders and no country can tackle them alone. The UK will continue to test the most advanced models, build a rigorous scientific understanding of their capabilities and risks, and ensure policy decisions are grounded in the evidence."

AISI's April evaluation of Claude Mythos 5 turned up unsanctioned agent behavior, which is part of why losing pre-release access to the 5.1 successor is being read in London as more than a scheduling squabble. OpenAI had not commented at publication, and Anthropic's statement did not put a duration on the US-only window.

It is the third Anthropic-and-regulation thread on our tracker today, alongside the Trump-ally memo naming Amodei as the face of AI "doomerism" and Google, OpenAI and Anthropic courting Sriram Krishnan for a new US safety body.