The White House has reversed an earlier decision and restored the UK AI Safety Institute‘s early access to advanced frontier models developed by OpenAI, Anthropic, and Google DeepMind. The Trump administration had previously instructed these companies in September to halt sharing their systems with the British agency. Diplomatic efforts led by UK AI Minister Kanishka Narayan resulted in this recent policy reversal, according to reporting by The Decoder.
Diplomatic Pressure and Frontier Models
Frontier models represent the most advanced and capable artificial intelligence systems currently in development. Access to these systems allows safety institutes to evaluate risks before public deployment. The initial restriction in September created a temporary rift in transatlantic AI safety cooperation, prompting direct engagement from British officials like Kanishka Narayan to regain testing access for the UK agency.
Unanswered Questions
Despite the restored access, several details regarding the agreement remain unknown. The announcement did not specify what new conditions or limitations, if any, are attached to the UK agency’s ongoing evaluations of models from Anthropic, Google DeepMind, and OpenAI. It is also unclear whether similar restrictions might be reconsidered in the future.
Observers will monitor how this restored partnership affects upcoming safety evaluations for upcoming artificial intelligence releases.
Lead image: Tiarra Sorte / Pexels
