US Government Forces Anthropic to Shut Down Fable 5 and Mythos 5: The First AI Kill-Switch

In an unprecedented move that sent shockwaves through the global AI industry, the United States government forced Anthropic to disable its two most advanced artificial intelligence models — Fable 5 and Mythos 5 — using an obscure export control directive that barred any foreign national from accessing the technology. The shutdown marks the first time a government has ever pulled a frontier AI model offline worldwide after public release, and it is reshaping conversations about AI sovereignty, national security, and the future of the open AI ecosystem.

Key Takeaways

• First-ever frontier AI model shutdown: On June 12, 2026, the U.S. Commerce Department ordered Anthropic to suspend access to Fable 5 and Mythos 5 for any foreign national — effectively pulling both models offline globally.
• Amazon researchers triggered the crisis: Amazon, a major Anthropic investor, reportedly jailbroke Fable 5 and demonstrated it could extract cyberattack-worthy information, then alerted the government.
• Global allies react with alarm: Leaders from the UK, Canada, France, and the EU condemned the move as a "kill-switch" proving overreliance on U.S. AI technology is a strategic vulnerability.
• Anthropic's $965B IPO under threat: The shutdown comes just as Anthropic confidentially filed for what could be one of the largest tech IPOs in history.
• Geopolitical ripple effects: France, Canada, and EU nations are accelerating plans for sovereign AI infrastructure and domestic models.

What Actually Happened: The Timeline

The sequence of events unfolded rapidly over just four days — from a model launch to a full global shutdown.

April 2026 — Mythos 5 Debuts, But Not for Everyone

Anthropic first unveiled Mythos 5 in April 2026 as a frontier AI model with extraordinary cybersecurity capabilities. The company itself described it as "too powerful" for public release, limiting access to a small group of vetted organizations through Project Glasswing, a cybersecurity initiative. Mythos 5 could identify severe code flaws and had advanced capabilities in vulnerability research that went beyond any publicly available model at the time.

June 9, 2026 — Fable 5 Launches to the Public

Anthropic released Fable 5 on June 9 — a public-facing version of Mythos 5 with additional safety layers. It was the first time Anthropic brought a model of this capability to hundreds of millions of users. The company claimed Fable 5 achieved state-of-the-art results on nearly all tested benchmarks. Just three days later, everything changed.

June 12, 2026 — The Kill-Switch Is Triggered

At 5:21 PM Eastern Time, the U.S. Commerce Department sent Anthropic an export control directive invoking national security authorities. The order required Anthropic to immediately suspend all access to Fable 5 and Mythos 5 by "any foreign national, whether inside or outside the United States, including foreign national Anthropic employees." Because it was technically impossible to geofence by citizenship, Anthropic had no choice but to disable both models for all users worldwide.

Why Did the Government Act?

The U.S. government's stated rationale centers on a jailbreak discovered by Amazon researchers. According to reports from The Wall Street Journal, Amazon — a major Anthropic investor with a $4 billion stake — found a way to bypass Fable 5's safeguards and extract information that could aid cyberattacks. Amazon notified the Commerce Department, which then acted.

Trump adviser David Sacks publicly stated that the administration had warned Anthropic about the jailbreak and that CEO Dario Amodei "refused" to patch or pull the model before the government imposed controls.

Anthropic disputes this characterization, arguing the jailbreak was narrow — only unlocking one specific cybersecurity capability — and that the same technique works on other frontier models like OpenAI's GPT-5.5, which were not subject to similar restrictions.

Global Backlash: Allies Scramble for AI Sovereignty

The shutdown triggered an immediate and furious reaction from U.S. allies around the world, who saw it as proof that American AI cannot be relied upon.

United Kingdom

Al Carns, UK former Minister for the Armed Forces and Labour MP, wrote on X: "This week the most advanced AI model on the planet got switched off by a foreign government. British researchers were studying it. British companies were testing it. British hospitals were piloting it. Not any more. This isn't an AI story — it's the story of every industry we depend on."

Canada

Canadian Prime Minister Mark Carney said: "Nobody has done anything wrong in the situation. We will have done something wrong if we just accept this, don't take the lesson, don't build out and diversify."

European Union

French MEP Christophe Grudler declared: "The United States is once again demonstrating what we have warned about so many times — that the US holds a real 'kill-switch' over essential technologies and that they are more than willing to use it." Dutch MEP Bart Groothuis added: "Europe needs its own LLMs and open weight models or face digital colonization."

France

President Emmanuel Macron announced a joint French-Indian AI initiative, emphasizing the need for "a reliable, open and safe AI that could be trusted." France is also replacing U.S. data analytics firm Palantir with domestic alternatives for government operations and transitioning away from American messaging apps.

What This Means for Anthropic's IPO

The timing of the shutdown could be devastating for Anthropic's financial ambitions. The company confidentially filed for an IPO in early June 2026 after a funding round that valued the company at $965 billion — making it one of the most valuable private companies in the world.

Investor enthusiasm for the IPO is now in question. The export control directive demonstrates that Anthropic's most advanced products can be taken offline at any moment by government fiat — a risk that is hard to price into an IPO valuation. Meanwhile, competitor SpaceX — which recently acquired Cursor AI for $60 billion — launched its own IPO on the same Friday, becoming the sixth-most-valuable public company in the U.S. with a $2.1 trillion market cap.

The Bigger Picture: A New Era of AI Geopolitics

This event is not happening in isolation. It represents an escalation in how governments view frontier AI models — not as software products, but as digital weaponry subject to the same export controls as advanced semiconductors and missile technology.

According to TechCrunch's analysis, the shutdown "should be a wake-up call for any U.S. tech company — AI lab or otherwise." The administration's action didn't appear to require court approval, raising constitutional questions about the executive branch's power over commercial AI products.

What Changes Now

Area	Before the Shutdown	After the Shutdown
Global AI trust	US models seen as reliable global infrastructure	Allies question reliability; seeking non-US alternatives
Model regulation	Voluntary commitments, no enforcement	Export controls used as de facto model regulation
Investor confidence	AI labs valued on model capability alone	Political risk now a major factor in AI valuations
Open-source AI	Growing but secondary to frontier labs	Accelerated push for open-weight models as sovereign alternative
US-China dynamics	Export controls focused on hardware (Nvidia GPUs)	Controls now extend to software model weights directly

Anthropic's Response and What Comes Next

Anthropic issued a public statement apologizing for the disruption and calling the government's action a "misunderstanding." The company argued that if the standard applied to Fable 5 were applied across the industry, "it would essentially halt all new model deployments for all frontier model providers."

Senior Anthropic staffers are meeting with Trump administration officials in Washington to try to resolve the dispute. Meanwhile, CEOs from Nvidia and Adobe are reportedly in talks with the administration to reinstate access, arguing the bans harm global cybersecurity efforts.

Security expert Alex Stamos told The Verge that companies are already rushing to sign backup contracts with non-U.S. providers of open-weight AI models. Even if access is restored, he warns, trust in the availability of U.S. frontier models has been permanently eroded.

What This Means for AI Users and Developers

If you're building on top of frontier AI models — whether through APIs, fine-tuning, or embedded agents — the Anthropic shutdown is a sovereign risk event. Your AI infrastructure can be unilaterally cut off by a government you may not even be subject to. The practical takeaways:

• Diversify your model providers. Don't build your entire stack on a single frontier model provider subject to U.S. export controls.
• Prefer open-weight models where possible. Llama 4, Mistral, DeepSeek, and other open models can't be remotely disabled.
• Plan for model volatility. Build abstraction layers that let you swap model providers without rewriting your application.
• Watch the regulatory landscape. The U.S. government has now demonstrated it will use export controls on AI models — this sets a precedent that could be applied to any frontier model from any U.S.-based provider.

Frequently Asked Questions

What exactly happened to Anthropic's Fable 5 and Mythos 5?

On June 12, 2026, the U.S. Commerce Department issued an export control directive ordering Anthropic to block foreign nationals from accessing both models. Since Anthropic couldn't verify user citizenship at the API level, the company disabled the models for all users worldwide.

Why did the US government target Anthropic specifically?

The administration cited national security concerns after Amazon researchers reportedly jailbroke Fable 5 and extracted information that could aid cyberattacks. The move also follows months of escalating tensions between Anthropic and the Trump administration over Pentagon contracts, AI safety practices, and what administration officials called Anthropic's "woke" regulatory capture strategy.

Can other countries still use US AI models?

Yes — this order specifically targeted Fable 5 and Mythos 5. Less powerful Anthropic models like Claude Opus 4.8 remain available, and models from OpenAI, Google, and other providers are unaffected — for now. However, the precedent has been set, and no US-based frontier model can be considered immune from future export controls.

How does this affect Anthropic's IPO plans?

Anthropic confidentially filed for an IPO in June 2026 at a $965 billion valuation. The shutdown introduces significant political risk that could dampen investor enthusiasm or force the company to delay its public offering. The full impact will depend on how quickly the dispute is resolved.

What should AI developers do in response?

Developers should diversify model providers, prefer open-weight models where feasible, build provider-agnostic abstraction layers, and monitor the regulatory landscape closely. The era of treating any single frontier model provider as a reliable infrastructure partner is over.

The Bottom Line

The shutdown of Fable 5 and Mythos 5 is the most significant government intervention in commercial AI to date. It marks the moment when AI model weights were formally reclassified from commercial software to strategic national assets — subject to the same export controls as nuclear technology and advanced weapons.

Whether you view this as necessary national security prudence or unprecedented government overreach, one thing is certain: the rules of the AI game have changed, and no one — not even the world's most advanced AI lab — is immune from the kill-switch.

What's your take on the US government's decision to pull Anthropic's models offline? Is this justified national security action or dangerous precedent-setting overreach? Share your thoughts in the comments below.