Anthropic Fable 5 Mythos 5 suspended
On Friday evening, the US government did something it had never done before. It ordered a company to switch off a live AI model that millions of people were already using.
The model was Fable 5, made by Anthropic, the company behind Claude. Its more powerful sibling, Mythos 5, went dark at the same time.
As of now, both are offline for everyone. Not just users in the US. Everyone, including here in the UAE.
What Actually Happened
Anthropic says it received a letter from the US Commerce Department at 5:21pm Eastern Time on June 12. The letter was an export control directive. It cited national security.
The order told Anthropic to block access to Fable 5 and Mythos 5 for any foreign national. That means anyone who is not a US citizen, whether they are inside the United States or outside it. It even covered Anthropic’s own foreign staff.
Here is the practical problem. A company cannot check the passport of every user in real time. So to follow the order, Anthropic had to take the simplest route. It turned both models off for all customers, worldwide.
Anthropic’s other models, including the regular Claude apps most people use, are not affected. Only these two top-tier models went dark.
Why the Government Stepped In
The trigger appears to be a jailbreak. A jailbreak is a trick that gets an AI model to do something its safety rules are meant to block.
According to reporting from Axios, another company claimed it had found a way to jailbreak Mythos. That claim reached the administration and raised alarm about cybersecurity risk. The government had reportedly already asked Anthropic to delay launching these models, and Anthropic went ahead anyway. The letter followed.
So this is the heart of the dispute. The government sees a powerful model with a possible security hole and wants it locked down. Anthropic sees a minor, narrow issue being used to justify pulling a product used by hundreds of millions of people.
Anthropic’s Side of the Story
Anthropic is complying with the order. But it disagrees with it, in public, and in plain terms.
The company says the jailbreak in question is narrow. Its understanding is that the technique essentially asks the model to read a piece of software and point out its flaws. Anthropic says that same capability is freely available in other public AI models, including OpenAI’s GPT-5.5, and that security defenders use it every day to find and fix problems.
It also says no tester has found a universal jailbreak, the kind that would broadly break the model’s safety controls. The issues raised so far, in its words, produced either harmless results or minor findings with no special uplift.
Anthropic’s bigger argument is about precedent. It says that if a single narrow vulnerability is enough to recall a live AI model, then by that standard no frontier model from any company could stay on the market. That would freeze the whole industry.
To be fair to the government, none of this has been tested in the open yet. Anthropic has reviewed the report it believes triggered the order, but the full national security reasoning has not been made public. We are hearing one side in detail and the other in summary. That is worth keeping in mind.
Why This Matters Beyond the Headline
Strip away the technical fight and one fact stands out. For the first time, a government has reached into the market and switched off a commercial AI model that was already running.
Until now, AI companies decided what to release and what to pull. This shows that a government can override that decision with a letter sent on a Friday afternoon. That is a real shift in who controls access to the most powerful AI tools.
For the AI industry, that is the story. Not the jailbreak. The precedent.
What It Means for the UAE
Here is the part that is easy to miss, and it is the part that should matter most to readers here.
Read the order again. It blocks access for any foreign national. If you live in the UAE and you are not a US citizen, that is you. You were not caught in the crossfire of a US-only rule. You were named.
In practice, very few people in the UAE will feel the loss today. Fable 5 was brand new. Mythos 5 was never widely available to the public. Most UAE residents and businesses use the standard Claude models, ChatGPT, Gemini, or Perplexity, and all of those keep working normally.
But the lesson underneath is bigger than one model. The most powerful AI systems are built in the US and treated by the US as national security assets. Access for the rest of the world can be switched off overnight, with no warning and no say from the user.
If you run a UAE business that depends on a single US AI model for something important, this is a quiet warning. Do not build your whole operation on one tool you do not control. Keep a backup. Know which model you would switch to if your main one disappeared on a Friday evening. That is no longer a far-fetched scenario. It just happened.
The Bottom Line
The US government forced Anthropic to take its two most powerful AI models offline. Anthropic complied, but says the security concern is minor and the move sets a dangerous precedent. The full reasoning on the government’s side is not yet public.
For most people in the UAE, daily AI use is unchanged. The models you rely on still work.
The real takeaway is about control. The best AI in the world sits behind a US national security gate, and you are on the outside of it. Plan accordingly. Keep your options open. And do not assume the tool you use today will be there tomorrow.
Anthropic says it believes this is a misunderstanding and is working to restore access. We will update this article as the story develops.
Robius.news — Dubai, UAE — 2026 | Built to be first. Built to be trusted.
