TL;DR

The White House has lifted its block on Britain’s AI Security Institute (AISI) getting new frontier models for safety testing, LBC reported on Monday 12 October, citing The Telegraph. Washington had asked Anthropic, OpenAI and other labs to keep their latest systems from the institute, and the reversal reportedly followed lobbying by AI minister Kanishka Narayan.

What changed

The original request had a visible cost. According to LBC, it was the reason Anthropic stopped UK officials from examining Claude Mythos 5.1, its newest model. LBC does not say whether Mythos 5.1 will now go to the institute for review.

LBC’s account does not say when access resumes, whether the reversal covers every lab or whether it comes with conditions, such as US agencies still testing first. Those details will decide how much ground AISI has actually recovered.

How the institute coped

Before the reversal, the institute had argued it could cope. Last month, replying to a request from Labour MP Liam Byrne, AISI head Henry de Zoete explained that nobody outside the US had seen Anthropic’s latest model before it launched, The Guardian reported. He maintained that his teams could still assess such systems once they were public, and said testing models both before and after release was core to the UK’s grasp of AI risk.

The episode also fed a wider argument about dependence. George Williamson, who now leads the Alan Turing Institute, told The Guardian last week that political and commercial pressures meant the UK “may not have access [to AI] in all the circumstances when we want it”, calling that “not a place where a nation wants to be”.

Looking forward

We reported in September that the US wanted first look at frontier models before AISI. The reversal removes that immediate obstacle, and it lands on the day Narayan was set to give a speech calling for AI firms to do more on safety. LBC notes the issue could remain a source of friction, given President Trump’s stated opposition to curbs on what he terms “super intelligence”. In our view, the episode shows how far UK pre-release testing depends on decisions made in Washington. A request made once can be made again, so firms relying on AISI evaluations should not treat restored access as settled.