Imagine spending three years helping build one of the most sophisticated artificial intelligence systems ever created. You wrote code for it, tested it, argued about its architecture in late-night Slack threads, and watched it surpass benchmarks that once seemed impossible. Then, on a Friday afternoon, you open your laptop and find you can no longer access it.
Not because you did anything wrong. Not because your role changed. But because of where you were born.
That is effectively what happened to a portion of Anthropic's workforce when the US Commerce Department issued an export control directive ordering the AI company to immediately suspend access to its two most advanced models — Fable 5 and Mythos 5 — for any foreign national, whether they were sitting in a San Francisco office or halfway around the world. The directive landed late on a Friday, as these things often do, and Anthropic found itself with a problem it could not cleanly solve: its platform had no mechanism to selectively block access by citizenship. So it did the only thing it could. It disabled the models for everyone, worldwide, until a compliant system could be built.
The story made headlines as a national security move. A cybersecurity directive. A regulatory adjustment in a fast-moving field. But read it more carefully and something else comes into focus. This is a story about who actually builds frontier AI — and what happens when the government's view of that question collides with Silicon Valley's.
