White House asks OpenAI, Anthropic to hold models from UK testers
Politico reports that a senior official wants US review to come first, and that Anthropic appears to have complied while OpenAI has not said what it will do.
The White House has asked OpenAI and Anthropic not to share new models with the UK government's testing agency until those models have been through US testing, Politico reported on Thursday. Sophia Cai, who wrote the story with Joe Bambridge, set out its findings on X.
Cai quoted a senior administration official on the reason. "Because they're American companies and this has been our policy with every new frontier model that comes out," the official told her. The official is not named in her account, and the exchange is the only quotation she gives.
The sequence the White House wants runs US review, then securing US systems, then sharing the model with US partners, according to Cai. Anthropic appears to have complied, she wrote. OpenAI has not said whether it will, and neither company has commented publicly on the request.
Samuel Hammond, a policy writer who reads on the US side of this argument, said he was not opposed in principle, but that the US institute should then be resourced at least as well as the UK's. No other outlet has matched the story so far.