White House asks OpenAI, Anthropic to hold models back from UK's AI Security Institute

The White House has asked OpenAI and Anthropic not to share new AI models with the UK's AI Security Institute (AISI) until the US government has tested them first, Politico's Sophia Cai and Joseph Bambridge reported on 24 September 2026, citing a senior administration official and a person familiar with the matter. A British official confirmed the report to Bloomberg the next day.

The request came from the Office of the National Cyber Director, part of the White House. It is a policy position, not a public rule, order or statute. OpenAI declined to comment to Bloomberg; Anthropic declined to comment to Politico; the White House did not respond to either outlet.

Anthropic appears to have gone along already, Politico reported. Claude Mythos 5.1, launched 1 September 2026, was restricted to a set of US organisations. Anthropic's own announcement said it was 'coordinating with the US government to expand access to a broader set of domestic and international partners as quickly as possible.' AISI was not given access to test the model before release.

AISI director Henry de Zoete addressed the gap in a letter to the House of Commons Business and Trade Committee on 15 September 2026: 'We maintain strong relationships with all frontier AI developers and continue to have prerelease access to some of the world's most capable models,' he wrote, pointing to its evaluation of OpenAI's GPT-6 Astra before its release. A UK government spokesperson told Politico: 'These risks do not stop at national borders and no country can tackle them alone.'

AISI reports £66m in annual funding, priority access to more than £1.5bn of compute and over 100 technical staff. Politico reported that the US Center for AI Standards and Innovation, the body Washington is asking labs to prioritise, has no permanent director and only a few dozen technical staff.

The request lands the same week Prime Minister Andy Burnham used the UN General Assembly to describe AISI as working 'hand in glove with the U.S. and the leading labs on AI safety' and called for 'a single set of global principles and standards'. President Trump, at the same UN meeting, rejected coordinated global AI governance as a 'globalist scheme'.

UK MPs had already been pressing the access question before this story broke. On 22 September, the Commons Business and Trade Committee summoned OpenAI, Anthropic, Google DeepMind and Meta to a hearing on 13 October, with AISI's de Zoete giving evidence the same day. Committee chair Liam Byrne said 'voluntary self-regulation is unlikely to be a sustainable safeguard for the future'.

Separately, this year's tension between Anthropic and the Trump administration predates this story: Trump ordered federal agencies to stop using Anthropic's products on 27 February 2026, and the Commerce Department briefly blocked foreign nationals from accessing Anthropic's Mythos 5 and Fable 5 models on 12 June 2026, before restoring access to some US organisations.