🇺🇸 🇬🇧The White House (白宫) reportedly requested US testing before OpenAI and Anthropic share new models with UK government testers.

According to Politico reporting relayed by Financial_Express, the White House (白宫) asked OpenAI and Anthropic not to share their latest models with UK government testers until US government testing is complete. The request reportedly originated with the Office of the National Cyber Director (美国国家网络总监办公室); the supplied material describes a request, not a published regulation or enacted legal prohibition.

The stated US objective was to assess frontier models and take measures to protect US systems before sharing them with allies. The UK AI Safety Institute (英国人工智能安全研究所), identified as AISI in the reporting, previously worked with the companies and typically received access before public model releases.

Anthropic reportedly had not yet provided Claude Mythos 5.1 to the UK institute. The company said it was coordinating with the US government to expand access gradually to domestic and international partners.

The same reporting says the US was strengthening the Commerce Department’s Center for AI Standards and Innovation (人工智能标准与创新中心), or CAISI. It also reports that OpenAI supported cooperation between CAISI and international institutions, including the UK institute, while UK Prime Minister Andy Burnham advocated international cooperation on AI safety.