White House Reverses AI-First Policy, Restoring UK Access to Unreleased Models

The Trump administration has reversed course on AI model access, permitting the UK's AI Security Institute to evaluate leading frontier models before their public release, marking a sudden shift in a "US-first" posture. The White House had imposed a block on September 24, directing OpenAI and Anthropic to withhold new models from the UK's testing body while Washington conducted its own national security review. The reversal, confirmed on October 11, restores what had been a cornerstone arrangement between the two governments since the 2023 AI Safety Summit.