White House restores model access for UK AI Security Institute
October 13, 2026
The White House has reversed a previous restriction, enabling the UK AI Security Institute to access unreleased frontier models. This includes testing capabilities from Anthropic, OpenAI, and Google DeepMind for safety evaluations.
HOW THIS AFFECTS YOU
●
policyYou can now oversee more rigorous, cross-border safety testing of the most advanced proprietary models.