[US Media: White House Requires New AI Models to Undergo U.S. Review Before UK Testing] According to a report by Jinse Finance on September 25, citing Politico, the White House has requested that OpenAI and Anthropic submit their latest AI models for U.S. government security review before granting testing access to the UK's Artificial Intelligence Safety Institute (AISI). According to informed sources, this request was made by the U.S. National Cyber Director's Office, as the White House wants the U.S. government to prioritize evaluating cutting-edge AI models to ensure the security of U.S. systems before sharing them with allies. The report states that Anthropic has not yet provided its Claude Mythos 5.1 model to the UK’s AISI and has indicated that it is coordinating with the U.S. government to gradually expand access to both domestic and international partners. The UK’s AISI has previously collaborated extensively with AI companies like OpenAI and Anthropic, often conducting safety tests before models are publicly released. This request comes as the U.S. government deliberates on how to regulate rapidly advancing AI models. Some AI companies and experts have previously warned that cutting-edge models, if not thoroughly tested, could pose safety risks. Meanwhile, the UK is advocating for the establishment of unified global AI safety standards. UK Prime Minister Andy Burnham has stated that AI safety requires international cooperation rather than being addressed by individual nations alone. The U.S. is also enhancing the capabilities of the Department of Commerce's National Institute of Standards and Technology (NIST) to make it a key agency for AI safety governance. OpenAI also announced this week that CAISI should collaborate with international organizations, including the UK’s AISI, to promote global AI safety standards.