On 12/06/2026, the US government ordered Anthropic to stop providing two models, Claude Fable 5 and Mythos 5, to foreign nationals. Because it was impossible to separate "foreign nationals" in real time among hundreds of millions of users, Anthropic took both models offline worldwide within just a few hours. This is considered the first time a live commercial AI model has been taken offline due to direct government intervention.
Quick summary
- When: 12/06/2026 (~5:21 PM US Eastern Time).
- Who ordered it: The US government, citing national security authority under export controls.
- Scope: Banned foreign nationals from using Fable 5 & Mythos 5 → Anthropic pulled both globally.
- Stated reason: A narrow "jailbreak" — asking the model to read source code and find/fix software vulnerabilities; only verbal evidence was provided.
- Lesson: Depending on a single foreign AI is a "can be revoked at any time" risk.
What happened?
According to Anthropic's official statement and major outlets (Al Jazeera, Fortune), at around 5:21 PM US Eastern Time on 12/06/2026, the US government issued an export-control directive requiring that foreign nationals be blocked from accessing Fable 5 and Mythos 5 — including Anthropic employees who are foreign nationals, whether inside or outside the United States.
Anthropic said there was no reliable way to classify foreign nationals in real time, so it was forced to disable both models for all customers worldwide in order to comply. Notably, Fable 5 had only just launched on 09/06/2026 — three days earlier.
Why was it banned?
According to Anthropic, the government's primary basis was a narrow "jailbreak" vulnerability: essentially asking the model to "read a specific codebase and fix the software bugs in it." Anthropic stressed that it only received verbal evidence, with no written technical documentation and no prior coordinated vulnerability disclosure process.
Anthropic also argued that this capability is inherent in many other models and is a routine operation for security engineers reviewing and patching software.
Anthropic's response
Anthropic stated that it would comply with the legal directive but disagreed that discovering a narrow jailbreak was enough to revoke a commercial model serving hundreds of millions of people. The company warned that applying this standard across the industry would "essentially halt all new model deployments from every frontier model provider."
Why this is a milestone
This appears to be the first time a leading AI company has had to take a publicly deployed commercial model offline due to a direct government order. It shows that access to a cloud AI model — even one you have paid for and depend on — can be revoked overnight for reasons beyond the business's control.
Lessons for Vietnamese businesses
For Vietnamese businesses building products or internal assistants entirely on top of a foreign cloud AI, the Fable 5 incident is a clear reminder:
- Availability is not in your hands. An administrative order in another country can cut off access instantly — even when you have done nothing wrong.
- Depending on a single provider is a single point of failure. When the model is pulled, every feature plugged into it stops with it.
- Sending data outside is a double risk: both a compliance risk (PDPL, cross-border data transfer) and a risk of losing access.
How to mitigate: multiple providers (multi-provider with fallback) and — for core data and processes — consider in-house AI running on the business's own infrastructure. When open-source models (Qwen, Gemma, etc.) run on-premise, no one can "turn it off" remotely, and the data never leaves the organization.
FAQ
Is Fable 5 banned permanently?
At the time of writing (22/06/2026), Anthropic said it was working to restore access and disagreed with the decision. The situation may change — this is for reference only, not a legal conclusion.
Are users in Vietnam affected?
Yes. The order targets "foreign nationals," and because Anthropic pulled both models worldwide to comply, users outside the United States — including Vietnam — cannot access Fable 5 & Mythos 5.
How can a business keep its AI from being turned off remotely?
Run open-source models on the business's own internal (on-premise) infrastructure. That way, both operational control and data stay in your hands, independent of a foreign provider's decisions.
Don't let your AI be turned off remotely
Namtech deploys private in-house AI platforms — open-source models running 100% on your infrastructure, with data on-premise and no dependence on foreign providers.
Book a free consultationNote: This article is compiled from public sources as of 22/06/2026; the situation may change. For reference only, not legal advice.