US Reverses Course on Britain's Testing of AI Models: AISI Regains Pre-Release Access

What happened

The Telegraph reported on October 11 that the Trump administration has reversed a decision that requested US AI developers not to share new models with Britain's AI Security Institute before release. Developers such as OpenAI and Anthropic can again give the British government early access to AI models for safety testing, Reuters reported the same day, citing the newspaper.

According to reports, the restriction began on September 24, when US officials asked developers to hold new models back from British testers until Washington finished its own national security review; the pause lasted about 17 days. During the pause, Anthropic's Claude Mythos 5.1 was made available only to US organizations. UK AI minister Kanishka Narayan and other ministers then lobbied Washington for the reversal, according to the Telegraph's reporting.

Spokespersons for Trump, OpenAI and Anthropic did not immediately respond to a request for comment from Reuters. Britain's AI Security Institute was set up after the world's first AI safety summit in 2023 and tests the most advanced AI models to evaluate their capabilities and the risk of misuse.

Key facts

1. Timing: The Telegraph reported the reversal on Sunday, October 11; Reuters carried the report the same day.
2. Who is affected: Britain's AI Security Institute (AISI) regains pre-release evaluation access to unreleased models from developers including OpenAI, Anthropic and Google DeepMind.
3. Length of the pause: The restriction reportedly began on September 24 and lasted about 17 days; during it, Claude Mythos 5.1 was available only to US organizations.
4. Responses: Spokespersons for Trump, OpenAI and Anthropic did not immediately respond to Reuters' requests for comment.