Three days after launch, Washington pulls Fable 5 — and the precedent is the story

The most consequential AI story of the week is not a capability. It is a mechanism. On 12 June 2026, three days after making Fable 5 public, Anthropic switched it off — along with the restricted Mythos 5 — to comply with a US government directive that barred access by any foreign national.

Read that scope again. Any foreign national — including Anthropic's own employees who are not US citizens. Because the order reached foreign nationals everywhere, the only way to comply was to turn the models off for everyone. Every other Anthropic model is unaffected; this was surgical to the Mythos lineage.

Here the reporting and Anthropic's account diverge, and honesty requires marking it as developing. Anthropic says it received only verbal notice of a 'potential narrow, non-universal jailbreak,' reviewed a demonstration that surfaced a few previously-known minor vulnerabilities, and disagrees that any of it warrants a recall. It also notes that rival models — it names OpenAI's GPT-5.5 — share the same underlying capability. The government's detailed rationale has not been published.

Anthropic is reportedly weeks from a major IPO, which sharpens every incentive to resolve this quickly and quietly. Watch three things: whether access is restored and on what terms, whether the government publishes its rationale, and whether the same theory is applied to rivals Anthropic says share the capability. Until then, the safest summary is the honest one — this is unresolved.