America’s attempt to calm fears of an AI kill switch may be giving foreign governments another reason to build one of their own. U.S. Secretary of State Marco Rubio has reportedly told American diplomats to convince overseas governments and companies that Washington cannot arbitrarily cut them off from U.S. technology, according to a diplomatic cable reviewed by Reuters. The cable instructed officials to challenge claims that temporary security restrictions amount to a kill switch and to push back against digital sovereignty programs that favor local infrastructure over American platforms.
The talking points argued that temporary pauses for security testing should not be treated as evidence of a kill switch. They urged diplomats to defend American platforms and challenge foreign digital sovereignty programs that restrict U.S. companies or require local infrastructure. The State Department has declined to comment on the cable, Reuters reported.
In essence, the secretary of state wants diplomats to reassure foreign buyers about American technology. The recent Anthropic cutoff of Fable and Samsung’s reported interest in Mistral shows why foreign governments and businesses are increasingly skeptical.
The Impact of Model Shut Offs
On June 12, the U.S. government barred non-U.S. residents from using Anthropic’s most advanced Mythos and Fable models on national security grounds. Anthropic responded by suspending access globally, since it could not immediately verify every user’s nationality. The restriction lasted only a matter of weeks, but its political effect was larger. Asian technology executives promoted local alternatives with recent model releases competing convincingly against OpenAI and Anthropic’s best models, and European lawmakers renewed calls for greater independence from American platforms.
The U.S. State Department’s efforts try to separate narrow security restrictions from the idea that American officials can switch foreign customers off at will. But the damage might already be done. A chief information officer choosing an AI provider does not need proof of a permanent kill switch. An abrupt, government ordered interruption may be enough. Banks, defense agencies, manufacturers and health systems cannot treat model access like a consumer streaming subscription. Once AI sits inside research, logistics or customer operations, continuity becomes part of the product.
Capital Shifting to Open and Sovereign Models
As an indication of the shift to open models, Samsung is discussing an investment in French AI company Mistral as part of a financing round that could value the startup at roughly €20 billion, Reuters reported, citing the Financial Times. Reuters described the possible investment as hundreds of millions of euros, but the Financial Times placed it near €1 billion. This is all preliminary, however, and hasn’t been publicly announced or verified.
Mistral offers a credible alternative. It develops European models, supports private deployment and has become a central piece of France’s effort to build domestic AI capacity. Microsoft recently agreed to spend billions of dollars on Mistral’s European computing infrastructure, letting Azure customers use the French company’s data centers and giving regulated industries another route beyond infrastructure controlled solely from the United States.
While the financing is not proof that Anthropic’s cutoff caused Samsung to seek a stake in Mistral, they expose the same strategic calculation. European and Asian companies want access to advanced models that are less vulnerable to political decisions made in Washington.
The Reassurance Campaign Could Backfire
In addition to attempting to reassure foreign customers and governments, the cable reviewed by Reuters also asks American diplomats to resist foreign programs built around digital sovereignty, including localization rules and policies that make it harder for U.S. technology companies to serve overseas markets.
This is causing yet more reason to hesitate when putting U.S. company’s models in the middle of their most important AI systems. American officials are asserting the right to restrict advanced technology for national security reasons. At the same time, they are asking foreign governments not to protect themselves against the consequences of those restrictions.
For sure, foreign governments and companies see the contradiction clearly. If advanced AI is sensitive enough for Washington to control, it may be too sensitive for another country to rent entirely from Washington. This continues to put AI in the center of geopolitical posturing and strategy.
