The US government issued an export control directive tonight ordering Anthropic to suspend all access to its two most advanced AI models — Fable 5 and Mythos 5 — for any foreign national, anywhere in the world. Because Anthropic cannot reliably separate foreign nationals from its broader user base in real time, the company disabled both models for every customer worldwide. All other Claude models remain unaffected.
Anthropic received the directive at 5:21pm ET on Friday. The stated concern: the government believes someone has found a way to jailbreak Fable 5. Anthropic disputes the severity of the claim, calling it a likely misunderstanding involving a narrow, non-universal vulnerability, but is complying while working to restore access. Commerce Secretary Howard Lutnick sent the letter directly to Anthropic CEO Dario Amodei.
Both models had launched just days earlier and were described by Anthropic as the most powerful AI systems it had ever released. That power, it turns out, is exactly the problem. The US government doesn’t invoke national security export controls — the same legal framework used to restrict nuclear and missile technology — over ordinary software. The fact that Fable 5 and Mythos 5 just landed in that category tells you everything about where AI development actually stands right now.
And somewhere tonight, a vibe programmer is staring at a half-built app and a blank prompt box, slowly realizing that maybe understanding what your code actually does was never really optional.
