The US government just killed the most powerful coding model 72 hours after launch

Claude Fable 5 shipped June 9. It was the most capable coding agent model on the planet.

By June 12 at 5:21 PM ET, the US government ordered Anthropic to shut it down for every user on Earth.

Not just foreigners. Every user. Anthropic had no way to verify nationality in real time, so compliance meant a full global disablement of Fable 5 and Mythos 5.

Here's what triggered it.

Commerce Secretary Howard Lutnick sent a letter to Dario Amodei citing national security. The government claimed a rival company had "jailbroken" Mythos 5 — bypassing its cybersecurity safeguards.

That rival? Amazon researchers. The "jailbreak" was a prompt that said "fix this code." The model identified known vulnerabilities in a codebase. That's what penetration testing tools have done for a decade.

Anthropic's response is scathing: "We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well."

The government's letter contained zero technical evidence. No specifics. No CVEs. No benchmark numbers. Just a verbal claim.

This is the first time the US government has retroactively banned a commercially available AI model through export controls. Not model weights. Not chips. An API endpoint.

The blast radius is immediate and structural:

Every agent workflow built on Fable 5 just broke. The $10/$50 per million token model that outperformed Opus 4.8 on cybersecurity tasks — gone overnight. No replacement path. No migration guide.

Every team that signed up between June 9 and June 12 for the Mythos-class advantage — dead investment. Anthropic is offering refunds.

Every enterprise that evaluated Fable 5 for autonomous security agents — your vendor just proved that model availability is a compliance surface.

Claude Opus 4.8, Sonnet 4.6, and Haiku 4.5 are unaffected. But Fable 5 was the reason teams upgraded. It scored higher on code review, autonomous vulnerability discovery, and long-horizon debugging than anything else Anthropic shipped.

As of June 15, Anthropic's senior technical staff are in Washington meeting with White House and Commerce Department officials. Fortune revealed the specific prompt behind the ban. Axios reports the dispute goes deeper — alleging Anthropic allowed Mythos 5 access to an entity with CCP ties.

No restoration timeline. No new model announcement.

If you built agent pipelines on Fable 5, reroute to Opus 4.8 immediately. Audit any tool that calls `claude-fable-5` by model ID. Those requests will fail silently or return 403. This is not a temporary outage — it's a government directive with no expiration date.

The era of "build on any frontier model" just ended. Model availability is now a geopolitical variable. Plan accordingly.