The White House has asked OpenAI and Anthropic not to share their AI models with the UK government’s testing agency until the models have gone through US testing, reports Politico

The mechanism here is jurisdictional primacy rather than prohibition, a demand that US evaluation precede foreign evaluation, which signals the administration views pre-deployment testing as a sovereign chokepoint. Episodes of this kind have tended to evolve from informal requests into formalised policy, so the tell is whether this stays a one-off ask or hardens into a standing requirement on developers with government contracts. The actors matter: both named labs have cultivated close White House relationships and have prior form in accommodating administration preferences, which raises the odds of quiet compliance over public pushback. Worth watching is the UK response, since allied testing bodies have been building parallel evaluation capacity and any friction there would speak to fragmentation of the AI governance landscape. For the listed names, the transmission runs through the majors' exposure to these labs rather than any direct operational effect.