Washington drops its 17-day request that OpenAI and Anthropic withhold models from UK testers
Washington lifted a 17-day request that OpenAI and Anthropic hold back pre-release models from the UK's AI Security Institute. The hold was set to last until a US national security review finished, and the reports do not say whether it has.
The Investor · Invest desk
What happened
- Crypto Briefing reported that the administration made its request on September 24, 2026, naming OpenAI and Anthropic.
- The Telegraph reported on Sunday, October 11, that US developers can now give the British government early access for safety testing without running afoul of the administration.
- Crypto Briefing reported that UK AI minister Kanishka Narayan lobbied for the reversal.
Why it matters
- What it limits constraint Because the UK's early access came from voluntary arrangements, a government request was enough to pause it.
- What it costs cost Anthropic released Claude Mythos 5.1 to US organizations only during the pause, so organizations outside the US were left out of its first distribution.
- What doesn't add up contradiction Reuters says Trump has repeatedly dismissed binding safety requirements for developers, yet his new AI task force said on Friday that developers must report security incidents.
Next story · Product · 3d ago
Two of four frontier AI labs tell UK MPs they want binding rulesConfirmed2 publishers
The count holds up. Six days remained in September after the September 24 request [1] and 11 days of October passed before the reversal [3], a total of 17 [10] that matches the figure Crypto Briefing reported [9]. The sources differ on what kind of instrument imposed the pause. Crypto Briefing's text calls the action a directive [15] and also says the administration asked the two labs to stop [1]. The Telegraph, as relayed by Reuters, describes a decision that requested developers not to share new models before release [4].
Most of the detail rests on one outlet. The September 24 date, the minister's lobbying and the three-lab list come only from Crypto Briefing [1][6][3], while the Reuters summary of the Telegraph report omits all three and notes that spokespersons for Trump, OpenAI and Anthropic did not immediately respond [18].
The three-lab list holds a puzzle. Google DeepMind was not named in the September request [11], yet it appears among the labs whose access is restored [3].
In our view the short hold is weak evidence that the arrangement is stable. The hold was short, or rather, short against an end condition that had no date on it [2], and if the review is still open, access came back with that condition unmet. The public signals are the Telegraph's account that developers can share without running afoul of the administration [5] and Reuters' description of a softer stance [14]. Crypto Briefing names whether the arrangement is formalized or stays voluntary as the thing to watch [19].
One path is a written framework; Crypto Briefing says one would give labs and allied governments more predictability [19]. The review could also end with a new hold. Or nothing is written, and the next pause arrives as this one did, by request.
The counter-thesis is that a hold undone in 17 days [10] is harder to reimpose than it looks. We would be wrong if the review concludes and the UK keeps access without a second request.
What to watch
- Whether the White House, OpenAI or Anthropic confirms the September 24 request and its scope; none had responded to Reuters when it published.
- Whether Anthropic widens Claude Mythos 5.1 beyond US organizations now that UK access is back.
- How the AI task force defines and enforces its incident-reporting requirement for developers.
Information, not investment advice. This story is general information about companies and markets, drawn from cited reporting. It is not a recommendation to buy, sell or hold any security; check the cited sources before acting.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence62
- Adoption
- Not scored
- Hype gap+12
- Incentives45
- Confidence66
Perspective Coverage
7 publishers- Builder
- Builder 27%
- Operator
- Operator 56%
- Investor
- Investor 17%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
On September 24, 2026, the Trump administration asked OpenAI and Anthropic to stop sharing new AI models with the UK's AI Security Institute.
- [2]
The hold was meant to last until a comprehensive US national security review was finished.
- [3]
As of October 11, 2026, the UK's AI Security Institute again has access to pre-release systems from Anthropic, Google DeepMind and OpenAI.
- [4]
The Telegraph said on Sunday that the Trump administration has reversed a decision that requested developers such as OpenAI and Anthropic not to share new AI models with Britain's AI Security Institute before release.
- [5]
American AI developers can give the British government early access to AI models for safety testing without running afoul of the Trump administration, the Telegraph said.
- [6]
UK AI minister Kanishka Narayan lobbied to reverse the decision.
- [7]
Anthropic released its Claude Mythos 5.1 model exclusively to US organizations during the interim period.
- [8]
Under earlier voluntary arrangements, the UK received early access to emerging AI systems so its researchers could probe them before the public did.
- [9]
Crypto Briefing reported that the pause lasted 17 days.
- [10]
September 24 to October 11, 2026 spans 17 days (6 days left in September plus 11 days of October).
- [11]
Google DeepMind was not named in the original September request, which targeted OpenAI and Anthropic specifically.
- [12]
Trump's newly launched AI task force said on Friday that developers must report security incidents with their AI models.
- [13]
Trump has repeatedly dismissed calls for binding safety requirements on AI developers, a position that appears to be shifting, Reuters said.
- [14]
Reuters described the change as signalling a softer stance by the Trump administration on international AI safety efforts.
- [15]
Crypto Briefing's text describes the US action as a directive that kept Britain from testing new AI models before their release.
- [16]
It is not clear whether the US national security review has concluded or what it found.
- [17]
It is unclear whether Anthropic will now expand Claude Mythos 5.1 beyond US organizations.
- [18]
Spokespersons for Trump, OpenAI and Anthropic did not immediately respond to a request for comment from Reuters.
- [19]
Crypto Briefing said to watch whether the restored arrangement gets formalized or stays voluntary, and that a written framework would offer labs and allied governments more predictability.
Sources
6 independent publishers whose own reporting we read for this story.
- channelnewsasia.comUS reverses course on Britain's testing of AI models, Telegraph says
2 articles · October 12, 2026
- cryptobriefing.comUS reverses course, restores UK access to frontier AI models for testing
3 articles · October 12, 2026
- economictimes.indiatimes.comAI models testing: US reverses course on Britain's testing of AI models - The Economic Times
1 article · October 11, 2026
- qz.comThe U.K. won back access to pre-release OpenAI and Anthropic models
1 article · October 12, 2026
- techrepublic.comUK Regains Early Access to AI Models After Reported White House Reversal
1 article · October 13, 2026
- the-decoder.comWhite House reverses course and restores UK AI Safety Institute's access to frontier models
1 article · October 12, 2026
- thenextweb.comUK says it will again test OpenAI and Anthropic models before release
1 article · October 12, 2026
Topics and entities
Topics
- Frontier model access controls
- AI safety testing