The US government has reversed a directive that temporarily blocked American companies from sharing their latest artificial intelligence models with the UK's AI Safety Institute (AISI). The decision restores the UK's ability to conduct pre-release safety tests on advanced AI systems from developers including OpenAI, Google DeepMind, and Anthropic.
The freeze began on September 24, 2026, and lasted 17 days while Washington conducted a national security review. UK officials lobbied against the halt, arguing it undermined Britain's role as an independent AI safety regulator.
Access was restored on October 11, 2026, signaling renewed US-UK cooperation on managing the risks of powerful new AI technologies.