US Bans Anthropic's Fable 5 and Mythos 5: First-Ever Export Control on AI Models

What Just Happened?

On Friday, June 12, 2026, Anthropic received an emergency export control directive from the US government: ban all foreign nationals from accessing Fable 5 and Mythos 5 — their two most powerful AI models just released.

Unable to enforce selective compliance, Anthropic was forced to disable both models globally.

This marks the first time in history the US government has applied export controls directly to an AI model — previously these measures were only used on semiconductor chips.

Why Were They Banned?

The US government discovered a potential jailbreak method that could bypass Fable 5's safeguards, enabling it to identify software vulnerabilities — capability deemed a national security risk.

The Commerce Department, under the Trump administration, activated an executive order signed 10 days earlier that established a framework for the government to assess the risk of advanced AI systems up to one month before public release.

Anthropic pushed back hard. In their official statement:

"We believe the government should have the ability to block unsafe deployments as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles."

Anthropic argued that the jailbreak in question is narrow and non-universal — capabilities that other models like OpenAI's GPT-5.5 also possess. They called it a "misunderstanding" and are working to restore access.

How Powerful Are Fable 5 and Mythos 5?

Fable 5 was released to the public on June 9. It's the first model in the Mythos-class tier — a new level above the previous Opus-class.

• Fable 5: Public release with strong safeguards. Anthropic said it excels at software vulnerability discovery.
• Mythos 5: Even more capable, fewer safeguards, initially limited to select partners under Project Glasswing.

Both are among the most capable AI models ever developed. Anthropic had warned that "releasing a model this capable comes with risks" and implemented a defense-in-depth strategy — including safeguards, monitoring, and 30-day data retention policy to detect jailbreaks.

The Strained Relationship Behind the Ban

This didn't happen in a vacuum. Anthropic and the Trump administration have been at odds for months:

1. Refusing the military: Anthropic wouldn't let the Pentagon use its models for autonomous weapons — only logistical support.

2. Pentagon blacklist: The DoD labeled Anthropic a "supply chain risk," banning Claude across the Pentagon.

3. Anthropic sued: They challenged the blacklist in court — still unresolved.

4. The ban: Now they're deemed "too dangerous for foreign use." The irony: refusing to serve the military → blacklisted → export banned.

Kirsten Davies, Pentagon CIO, posted on X: "Some things are simply more important than revenue cycles, clickbait, and pre-IPO valuation. America First. Always."

Impact and Implications

For Anthropic

The timing couldn't be worse. Anthropic is preparing for an IPO — losing global access to their most powerful models will seriously impact valuation.

Worse, this creates a perverse incentive: governments and enterprises worldwide now have more reason to build their own models or switch to Chinese open-source alternatives like DeepSeek and Qwen.

For the Industry

This sets a dangerous precedent. If a narrow, non-universal jailbreak is enough to recall a model, every frontier AI company could be affected.

As one analyst put it: "When you spend years describing your model as potentially civilization-ending, don't be surprised when governments start treating it like weapons-grade material."

For Users

If you're using Fable 5 or Mythos 5 via the Claude API — you're locked out until Anthropic restores access. Other Anthropic models remain unaffected.

Technical Reality Check

Anthropic confirmed the jailbreak was essentially "asking the model to read a codebase and find bugs" — a capability available in other models including GPT-5.5, and used daily by security defenders worldwide.

The core issue here isn't new technical capability — it's political theater around AI safety.

The Bottom Line

This marks a turning point in US-AI industry relations. Export controls have expanded from chips to models themselves.

Anthropic is caught between serving global customers, complying with government directives, and preparing for an IPO. The outcome of this battle will affect the entire industry — and your access to the world's most advanced AI models.

Anthropic says they're "working to restore access as soon as possible." Stay tuned.