AI Trend Research

← October 12, 2026 briefing · added in the 16:00 KST update

P1 Policy Unconfirmed Safety & alignmentPublic sector16:00 updateUS administrationOpenAIAnthropic

Trump administration reverses course, lets US AI firms give UK AI Security Institute pre-release model access (report)

Announced October 11, 2026

Verified (fact)

AI Times, citing a Telegraph report from Oct. 11 (local time), said the Trump administration has reversed its earlier position that told US companies such as OpenAI and Anthropic not to share unreleased AI models with the UK AI Security Institute (UK AISI). US AI developers can now reportedly give new models to UK AISI for safety evaluation before official release without worrying about penalties. The source material does not include an official US government announcement.

Why it matters (interpretation)

This reopens a path for national-level safety evaluation of US frontier models before release. Read together with the White House's recent mandate to report AI security incidents, it can be seen as a sign that the US is moving back toward accepting external safety evaluation (interpretation). It could also affect safety evaluation cooperation with the UK and Japan.

Confidence medium · official source pending

Sources

More about US administration