The White House lets Britain's AI Security Institute test new OpenAI and Anthropic models before release again, the Telegraph reports, and London says testing is under way
The Telegraph reports that a freeze ordered by the US cyber office has been lifted. The UK government confirms the evaluations but not the reversal.
The Trump administration has reversed the directive that kept Britain's AI Security Institute from testing unreleased frontier models, the Telegraph reported on 11 October, according to Startup Fortune and The Next Web. According to Startup Fortune, the White House Office of the National Cyber Director ordered the pause on 24 September, pending a US national-security review, and lifted it on 11 October, restoring the institute's pre-release access to models from Anthropic, Google DeepMind and OpenAI. The Telegraph said AI minister Kanishka Narayan had lobbied for the change.
On 12 October the UK government said scientists at the institute “are now conducting rigorous evaluations” of the latest generation of advanced models, following work with leading AI developers and the US Government. That release carried Narayan's message to frontier AI developers, which the Switchboard covered on 13 October; the government page does not mention a US pause or a reversal.
Narayan said the UK had worked closely with the US to keep that access, Bloomberg reported, according to The Next Web. “When you're reliant on someone else's technology, you have to be clear-eyed about the permanence of any outcome,” he told Bloomberg. The Next Web says he was “very confident” the UK could test the latest models. Anthropic did not share its Mythos model with the institute, Bloomberg reported, according to The Next Web. The White House, OpenAI and Anthropic did not comment on the record in the coverage read.
- Reported The White House lifted its freeze on UK pre-release testing, the Telegraph reported, according to Startup Fortune. Startup Fortune
- Confirmed The UK government says AI Security Institute scientists are now conducting rigorous evaluations of the latest models, after work with developers and the US Government. GOV.UK
- Reported Narayan said he was “very confident” the UK would be able to test the latest models, The Next Web reported. The Next Web
Safety, security & governanceCourts & policy
Caught up on October 17, 2026, after it was missed at the time · front page