Anthropic restores Fable 5 after 18‑day US export control shutdown

Fable 5’s 18‑day disappearance and quiet return is a preview of what frontier releases will look like in a world where export controls and national‑security review are baked into the launch process. Anthropic didn’t hit a technical wall; it hit a policy wall, with U.S. authorities effectively forcing a global pause to reassess how and where its most capable model could be used.

In the race to AGI, that’s a subtle but important shift. For years, the default assumption was that once a lab could train a model, it could deploy it globally, subject mostly to its own risk calculus. Now we’re in a regime where state actors can retroactively yank a model from production, and then selectively allow it back with conditions. That creates a new “release test” alongside evals and red‑teaming: can your launch survive government scrutiny without being shut down?

Practically, this will drive labs to design tiered model families (like Fable vs Mythos) with different risk profiles and exportability baked in. It could slow the public rollout of the most powerful systems even as internal capabilities continue to climb, nudging the effective timeline for broadly accessible AGI‑class models later than the raw research curve would suggest.