On June 12, 2026, the US government issued an export-control directive ordering Anthropic to block Claude Fable 5 and Mythos 5 from all foreign nationals. Rather than partially comply, Anthropic disabled both models for every customer worldwide — just three days after Fable 5's public launch. All other Claude models remain unaffected.
What Did the Government Actually Order?
The directive reached Anthropic at 5:21 PM ET on June 12, 2026, in a letter from Commerce Secretary Howard Lutnick to CEO Dario Amodei. Citing national security authorities, it placed Claude Fable 5 and Mythos 5 under export controls: any export, re-export, or domestic transfer of the models to a non-US person would now require government approval first. Critically, the restriction applied not just to people outside the country, but to every foreign national inside it — including Anthropic's own non-citizen employees.
Faced with that scope, Anthropic said selective compliance wasn't practical: honoring the order precisely would have meant walling off a large share of its own users and staff. So the company chose to shut the two models down entirely, for everyone. Fable 5 had launched only days earlier as Anthropic's first Mythos-class model available to the public, priced at $10 per million input tokens and $50 per million output tokens and offered across Amazon Web Services, Google Cloud, and Microsoft Foundry. Microsoft had already temporarily barred its own employees from using it — over a separate worry tied to a new 30-day data-retention requirement.
Why Did Regulators Move So Fast?
The government's stated concern was security. According to Anthropic, the directive followed a claim from another company that it had found a way to "jailbreak" Fable 5 — bypassing the safeguards meant to block dangerous outputs in areas like cybersecurity and biology. But the letter itself, Anthropic says, included no specific technical detail of the national security concern, and the only evidence shared with the company was verbal.
This is the first time I've seen the US use export-control authority to pull a commercial model that was already serving people at scale. Whatever you make of the security claim, the bigger story is the precedent: frontier AI is now being governed through trade law — sometimes faster than the public can see the evidence. For anyone building a business on top of a single model, that's a new kind of supply risk. Your core tool can disappear overnight by directive, not by outage.
Anthropic has pushed back firmly. It says the technique in question amounts to a narrow, "non-universal" jailbreak — essentially asking the model to read a codebase and fix software flaws — and that it surfaced only a few minor, already-known vulnerabilities. The company notes that other widely available models, including OpenAI's GPT-5.5, can do the same thing without any bypass at all. Before launch, Anthropic says, Fable 5 was red-teamed for thousands of hours alongside the US government, the UK's AI Safety Institute, and outside experts, and no universal jailbreak was ever found. The 30-day data-retention policy that drew complaints exists, it argues, precisely so the company can detect and shut down attacks quickly.
What Does This Mean for Users and the Industry?
For everyday users, the immediate effect is narrow but real: Fable 5 and Mythos 5 are gone for now, while Opus, Sonnet, and Haiku keep working normally. For companies that wired the new models into products in the days after launch, it's a sharper lesson in continuity risk. And for the industry, the episode raises a question that will outlast this one model. If a narrow, non-universal jailbreak is enough to recall a deployed frontier model, Anthropic warns, the same standard could effectively freeze new releases across every major lab.
Anthropic says it is complying with the order while disputing it, and that it believes the action stems from a misunderstanding it hopes to resolve quickly. The company has also argued — consistent with its public policy positions — that governments should be able to block genuinely unsafe deployments, but through a process that is transparent, fair, and grounded in technical facts. Its core complaint here is that this directive was not.
Frequently Asked Questions
Why did the US government ban Claude Fable 5 and Mythos 5?
The US Commerce Department issued an export-control directive on June 12, 2026, citing national security and a reported method of jailbreaking Fable 5. The order barred access for all foreign nationals, and Anthropic responded by disabling both models for every customer worldwide.
Is Claude Fable 5 still available?
No. As of June 12, 2026, Claude Fable 5 and Mythos 5 are disabled for all customers worldwide while Anthropic complies with the directive. Anthropic says it is contesting the order and hopes to restore access soon.
Are Claude Opus, Sonnet, and Haiku affected?
No. The directive applies only to Fable 5 and Mythos 5. All other Claude models — including Opus, Sonnet, and Haiku — remain fully available.
What was the jailbreak that triggered the ban?
According to Anthropic, it was a narrow, non-universal technique that asked the model to read a codebase and fix software flaws, surfacing only minor, already-known vulnerabilities. Anthropic says comparable capabilities are available from other models, including OpenAI's GPT-5.5.
When were Claude Fable 5 and Mythos 5 banned?
The directive reached Anthropic at 5:21 PM ET on June 12, 2026 — just three days after Fable 5's public launch on June 9, 2026.
The Fable 5 and Mythos 5 shutdown is less about one jailbreak than about who gets to pull a live AI model offline, and how. Anthropic complied within hours but is contesting the decision, calling it a likely misunderstanding. The takeaway for builders: frontier-model access is now a regulated — and revocable — resource.