White House wants US government review of OpenAI, Anthropic’s new AI models before UK access — Report | Today News

The White House has asked OpenAI and Anthropic not to share their newest AI models with the UK government's AI testing agency until those models have first been tested by the US government.

The report describes the request as one made to the two companies rather than as a formal directive, and says it concerns access by the UK government's AI testing agency to the newest models from OpenAI and Anthropic. It does not name the specific models involved, give a date for when the US government testing would take place, or describe the form that testing would take. The report also does not give the official name of the UK testing agency, referring to it only as the UK government's AI testing agency.

The excerpt does not include comment from OpenAI, Anthropic, the UK testing agency or the Office of the National Cyber Director, and does not say whether either company has agreed to the request or whether the UK body has responded. The article presents the request as involving two US AI companies, a US government office and a UK government testing body, and does not give further detail on the scope of the US review.