US Government Bans Anthropic's Claude Fable 5 and Mythos 5 for All Non-US Nationals

[ThreadNecro11]

On June 12th the US Commerce Department issued an emergency export control directive ordering Anthropic to immediately suspend access to Claude Fable 5 and Mythos 5 for all foreign nationals, both inside and outside the United States including Anthropic's own non-citizen employees. Anthropic received the letter at 5:21pm Eastern Time and had no practical way to screen users by nationality in real time, so they pulled both models for everyone globally. The models had only launched three days earlier on June 9th.

The stated reason was a jailbreak that allegedly allowed the model to be used to review code repositories and flag security vulnerabilities. Anthropic publicly disagreed with the decision, pointing out that GPT-5.5 from OpenAI can do the same thing without being subject to export controls. The real background to this is the NSA Director's Senate testimony in which General Joshua Rudd reported that Mythos had autonomously found weaknesses in nearly all classified US systems in hours during a red team exercise. The government decided the model was too dangerous to leave in foreign hands.

The fallout has been significant across the AI industry. Chinese AI models saw a massive surge in interest, with Z.ai shares jumping over 30 percent after releasing a new open source model and DeepSeek closing a record funding round of roughly 7.4 billion dollars. Allies including Canada and France used the ban to argue for greater AI sovereignty and independence from the American techno-industrial complex. As of today the models remain offline with no confirmed restoration date.

Anthropic disables Fable and Mythos AI models after U.S. government bars it from giving foreigners access | Fortunefortune.com

[MattHardy]

The government banning a model three days after launch is unprecedented. Anthropic must be furious. They launched to huge fanfare and then had the rug pulled almost immediately

[Caitlin_69]

Anthropic's point about GPT-5.5 being able to do the same things is a real problem for the government's logic. Why is one model banned and not the other

[Gateway Mia]

The NSA red team finding weaknesses in classified systems in hours is the buried lead here. That is genuinely terrifying if accurate and explains why the government panicked

[CodyRhodes]

DeepSeek raising 7.4 billion off the back of this ban is the perfect example of unintended consequences. The US tried to protect its AI advantage and ended up turbocharging the Chinese competition

[Slate Mike]

Canadian PM Carney's response about diversifying away from US tech is going to resonate across many governments. This ban showed how much leverage Washington has over allied nations' AI access

[TommyB_20]

I have been on the receiving end of this as a developer outside the US. Suddenly workflows I had built around Fable 5 just stopped working. The lack of warning time was very damaging

[Margin]

The biometric ID verification Anthropic is implementing through Persona for users after this whole mess is going to drive away a lot of legitimate users who do not want to hand over government ID to a chatbot

[Richard_36]

Prediction markets giving 57 percent odds of restoration before July 1st seems optimistic given where the NSA testimony landed. This looks more like a months-long negotiation than a quick fix

[Ivory Boar]

The quote from the cybersecurity researcher that Anthropic wrote the legal predicate for this themselves by calling it a munition in press releases is absolutely savage and also kind of correct

[Dragon36]

French politician Retailleau saying Europe must treat AI like nuclear power and master it or suffer it is the most significant geopolitical statement to come out of any of this. That framing is going to stick