The Trump administration just made history: for the first time, a US AI company has been cut off from serving customers abroad. Effective immediately, all foreign nationals — including Anthropic employees working in the US — are barred from accessing Mythos 5 and Fable 5. This isn't a voluntary restriction. This is an export control.
What happened
It started with a jailbreak. Amazon researchers discovered a vulnerability in Anthropic's Fable 5 model and reported it through proper channels. Instead of patching the hole, David Sacks says Anthropic CEO Dario Amodei refused to "fix the jailbreak or de-deploy the model." The Trump administration responded by imposing export controls that effectively lock Anthropic out of every market outside the US.
"The Trump administration's decision to impose export controls on Anthropic followed multiple tense calls between Dario Amodei and admin officials."
That's according to US officials cited in Politico. Senior Anthropic technical staff are now in DC trying to fix the dispute. Both sides say they're eager to resolve it — but the precedent is set.
Why it matters
Canadian PM Mark Carney put it plainly: the Anthropic ban shows the dangers of "over-reliance on certain models," and compared the risks to those that led to the 2008 financial crisis. That's not hyperbole. If a single company can be cut off from the global market on a government's whim, every non-US company has a reason to hedge.
Andy Jassy raised concerns with Trump officials about Mythos 5 — and that helped set the restrictions in motion. Amazon, one of Anthropic's biggest investors and partners, effectively helped trigger the very controls that now limit their investment's reach.
This is the AI version of "too big to fail" — except the failure here isn't financial. It's strategic. The US government just demonstrated it can weaponize model access the way it weaponizes chip exports. Every foreign company relying on US AI models now has a reason to rethink that dependency.
The knock-on effects
European political figures are calling the Mythos 5 ban a "wake-up call" about depending on the US for AI technology. Meanwhile, Apple's Siri team is already tapping third-party AI models beyond OpenAI in iOS 27 builds — a sign that even US companies are diversifying away from any single provider.
Satya Nadella put it best: companies must own their AI "learning loops" to compound human and token capital, or risk ceding all value to a handful of frontier models. The lesson just got more expensive.
The Anthropic crisis won't be the last of its kind. When a government can pull the plug on a company's global reach over a model vulnerability, everyone — including US companies — learns to build for a world where access isn't guaranteed.