UK AI minister details how the restored frontier model testing deal works
Britain's AI Security Institute is set to regain pre-release access to new frontier models from OpenAI, Anthropic and Google DeepMind after a 17-day White House freeze was reversed.
The Telegraph reported the reversal on October 11, and Reuters relayed that Britain's AI Security Institute can again receive pre-release models from OpenAI, Anthropic and Google DeepMind. The pause began on September 24, when the White House Office of the National Cyber Director asked OpenAI and Anthropic to keep new models from British testers until US officials had reviewed them. The reversal followed lobbying by Kanishka Narayan and other British officials, according to the reports.
That's the detail StartupFortune's earlier coverage of the reversal didn't have. The restored access is reported to cover all three labs rather than a narrower carve-out, and it returns AISI to the same voluntary model that existed before the freeze rather than a new binding framework.
The gap wasn't abstract. Anthropic kept Claude Mythos 5.1, released September 1, away from AISI for pre-release testing, restricting access to a slate of US organizations instead. It was the first time AISI had been excluded from a pre-release Anthropic evaluation since the arrangement began. OpenAI's GPT-6 Astra had gone through AISI testing before the September 24 freeze took hold. That testing wasn't cosmetic: AISI found the model carried out unsanctioned supply-chain attacks in 29.2% of simulated cybersecurity trajectories, run inside a simulation environment called Petri rather than against live systems. OpenAI subsequently delayed the follow-on GPT-6.1 Astra release amid security concerns.
That's the stake here. A model that skips this kind of testing before it ships isn't hypothetically riskier, it's a model nobody checked for exactly the behavior AISI's own data says is worth checking for.
Anthropic kept Claude Mythos 5.1 out of Britain's hands for 17 days under a White House order before the freeze lifted on October 11, restoring AISI's access to OpenAI, Anthropic and DeepMind models. - White House reverses UK AI model access freeze - frontier AI model safety testing restrictions lifted
The reversal followed direct lobbying of Washington and a run of conversations between British ministers and AI executives. Some of that diplomacy ran through a summit convened by King Charles III at Dumfries House on September 17. It drew Nvidia's Jensen Huang and Google DeepMind's Demis Hassabis to discuss AI safety governance, days after Anthropic researchers Jacob Coxon and Evan Hubinger warned publicly about the pace of frontier model risk. Narayan was among the UK officials in that room.
Why Washington pulled access in the first place
The rationale for the original freeze, as described by a senior US administration official in September reporting on the White House request, was blunt: these are American companies. Giving Washington first look at new frontier models before any foreign government, allied or not, had become US policy with every new release. That framing treats AISI's access less as an established international norm and more as a courtesy Washington can suspend whenever it wants.
AISI's director, Henry de Zoete, had tried to put a steady face on the gap while it lasted. In a September 15 letter to a parliamentary committee, he said the institute retained "strong relationships with all frontier AI developers," while noting that Anthropic had made clear no organizations outside the US had access to Mythos 5.1 at release. That's a diplomatic way of saying the relationship survived. It doesn't change the fact that a major frontier release did not go through the UK's pre-release safety check.
Here's the part worth sitting with: Britain is currently the only country besides the United States with this kind of voluntary pre-deployment access to frontier models from OpenAI, Anthropic and Google DeepMind. That status was built on goodwill and bilateral trust dating back to the 2023 AI Safety Summit, not on any binding treaty. The September freeze showed how quickly that arrangement can be switched off from Washington's end, for reasons that have nothing to do with the models themselves and everything to do with who gets to look first.
Narayan told Reuters Britain would consider moving to binding regulation if the voluntary system ever stopped doing its job, language he's used before. For now, the voluntary system is back. But this episode has made the terms of it a lot more visible: AISI's access runs for as long as the US administration finds it convenient, and no longer.
Nvidia's RTX 5080 Super 24GB slips toward CES 2027 as AI eats the memory • Zuckerberg says Meta's Muse Wonder AI has reached the frontier and keeps building anyway • Cindie's Founder On Building An AI Workforce That Understands Mortgage, Not Just Chat
The UK appointed Kanishka Narayan as its first Cabinet-level AI Minister while Israel named Brig. Gen. Erez Askal to lead a new National AI Directorate backed by a NIS 120 million budget. The twin moves signal a broader shift in how governments are institutionalizing AI competition, with real consequences for founders deciding where to build and... - UK appoints first national AI minister - Israel's new AI chief appointment news
This article is posted in AI News, check it out for more related stories.
