The White House has requested that OpenAI and Anthropic not present their new artificial intelligence models to the UK government testing agency until these models are reviewed and tested by the United States government. This request has been made by the Office of the National Cyber Director and places the companies in a difficult position.
Challenges for the Companies
This request pressures the companies to choose between not providing their models to the UK’s artificial intelligence security agency, which is one of the best government testing agencies in the world, and creating dissatisfaction within the Trump administration. A senior US government official stated that the White House wants the models to be reviewed by the United States first to ensure that American systems are secure.
Read more: The Conflict of Views Between the US and China on AI Regulation at the United Nations
Anthropic appears to have agreed to this request and has not currently provided its Claude Mythos 5.1 model to the UK’s artificial intelligence security agency. The company stated that this model is only available to a select group of American organizations.
Security Concerns and International Cooperation
In a letter to the UK parliamentary committee, Henry de Zoete, the director of the artificial intelligence security agency, referred to the lack of access to the Anthropic model and said, "We have strong relationships with all AI developers and still have pre-release access to some globally capable models." He also mentioned testing OpenAI's GPT-6 Astra model before its release.
Meanwhile, statements from senior officials and former engineers of major AI companies have warned that if artificial intelligence models are left unchecked, they could have catastrophic consequences. The UK government is also working to establish close cooperation with the world's leading AI laboratories to assist in testing new models.
The UK Foreign Secretary, Edward Miliband, also referenced this issue at the United Nations Security Council, stating that governments must ensure that new models are thoroughly tested and have sufficient visibility to assess the activities of companies.
Read more: The Increasing Use of Artificial Intelligence by Voters for Election Information · Xi Jinping Emphasized the Responsibility of China and the US in Controlling Artificial Intelligence
