This morning I sat down to try Fable 5, Anthropic's brand-new flagship Claude model, and got a flat error: this model is not available. I assumed it was a glitch on my end. A stale app, a region thing, a bad rollout. It wasn't. A few minutes of reading turned up the real reason, and it's a much bigger story than a broken button.
Anthropic had been ordered by the US government to switch Fable 5, and its sibling model Mythos 5, completely off. Not throttled, not region-locked. Off, for everyone on the planet. Here's exactly what happened, why even paying customers outside the US lost access, and what it means if, like me, you're a foreign national who was just starting to use it.
Prefer to watch? Here's the 90-second version:
What actually happened
According to Anthropic's own statement, the US government issued an export-control directive, citing national security authorities, ordering the company to suspend all access to Fable 5 and Mythos 5 for any foreign national, whether inside or outside the United States, including Anthropic's own foreign-national employees.
That's an almost impossible instruction to follow surgically. There's no reliable, real-time way to verify the nationality of every person hitting an API or a chat box. So Anthropic did the only thing that guaranteed compliance:
"The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance."
In other words, because they couldn't cleanly screen foreign nationals out, they switched the models off for everyone, US citizens included. That's why my error message looked identical to what a user in California would have seen.
The timeline: it happened fast
- June 9, 2026. Anthropic launches Claude Fable 5 and Mythos 5, rolling out across its own apps and partner platforms (GitHub Copilot, AWS, and others).
- June 12, 2026, 5:21 pm ET. Anthropic receives the government directive. By its own account, that's barely three days after launch.
- Within hours, the models are disabled for all users, everywhere. Partner platforms confirm the cutoff too; Fable 5 was pulled across all GitHub Copilot experiences as of June 12.
- June 13. I (and presumably a lot of other people) hit the "model not available" wall and go looking for answers.
Why the government pulled it: the jailbreak claim
The stated trigger was security. Per multiple reports, the government believed it had become aware of a method of "jailbreaking" the models, bypassing their safety guardrails, with particular concern around Mythos 5's cybersecurity capabilities.
Anthropic, notably, pushed back on how serious this actually was. In its statement, the company said it reviewed the technique and found it amounted to "a small number of previously known, minor vulnerabilities," the kind that "other publicly-available models are able to discover" anyway. It characterised the exploit as a narrow one, essentially asking the model to read a specific codebase and fix software flaws, rather than a universal key that defeats all of Fable 5's safeguards.
So you have a genuine disagreement in public view: the government treating it as a national-security-grade risk, and the model's own maker saying the concern is narrower than the blanket shutdown implies.
What this means if you're outside the US
This is the part that hit home for me, writing from Assam. The directive specifically targets foreign nationals, so people like me were the explicit reason for the restriction, even though the practical fallout landed on everyone. A few takeaways:
- It wasn't a you-problem. If you got the "model not available" error on Fable 5, nothing was wrong with your account, app, or region settings. The model was globally off.
- Geopolitics is now part of the AI stack. Which frontier model you can use may increasingly depend on your passport, not just your subscription. That's a new reality worth planning around if you build on these tools.
- Don't hard-wire your workflow to a single bleeding-edge model. A model that launched Tuesday and vanished Friday is a strong argument for keeping your work portable across models.
The good news: most of Claude still works
Crucially, this was not a Claude-wide shutdown. Anthropic was explicit that "access to all other Anthropic models will not be affected." That means the models most of us actually use day-to-day are untouched:
- Claude Opus 4.8: the current heavyweight, still fully available.
- Claude Sonnet 4.6: the balanced workhorse.
- Claude Haiku 4.5: the fast, cheap option.
So if you opened Claude today and just picked a different model, you'd never have noticed anything was wrong. Only Fable 5 and Mythos 5, the two newest and most powerful models, went dark.
Will Fable 5 come back?
Anthropic says it is "working to restore access as soon as possible" and promised to share more details. As I write this, there's no confirmed return date, and the path back depends on satisfying a government export-control order, which isn't something a company resolves with a quick patch. My honest read: expect this to take a while, and don't be surprised if access returns in a more restricted form (US-only, or with nationality verification baked in) rather than the open launch we briefly saw.
💡 If you were relying on Fable 5: switch to Claude Opus 4.8 for now. It's the most capable model still open to everyone, and for the vast majority of tasks you won't feel the difference.
The bottom line
Fable 5 didn't break, and you didn't do anything wrong. It launched on June 9, and three days later a US export-control directive aimed at keeping it out of foreign nationals' hands forced Anthropic to disable both Fable 5 and Mythos 5 for every user worldwide. The company disputes that the underlying security concern justifies the scope, and says it's trying to restore access. Until it does, Opus 4.8, Sonnet 4.6, and Haiku 4.5 carry on as normal. The whole episode is a sharp reminder that at the frontier, AI access is now a geopolitical question as much as a technical one.
I'll be covering how this develops, and testing the alternatives, on the Neel's World YouTube channel.