.svg)
On 9 June 2026, Anthropic launched Claude Fable 5 - the most powerful AI ever made available to the general public. On 12 June, Washington ordered its immediate suspension for all non-Americans.
3 days. That's all it took.
And while you read this article, Americans are using it. You are not. 👇
What was Fable 5?
Fable 5 is the first Mythos-class model that Anthropic has made available to the general public. It is the same underlying model as Mythos 5, but with safeguards enabled for cybersecurity, biology and chemistry. Launched on 9 June 2026, it was available on claude.ai, Claude Code and the API at $10 per million input tokens.
The first results were spectacular. Stripe reported that Fable 5 compressed months of engineering into a few days - completing in a single day a migration on a 50-million-line Ruby codebase that would have taken an entire team two months.
On financial analysis, code, vision and reasoning, it outperformed everything that existed. It was a genuine breakthrough.
But Anthropic is pushing back - and its arguments are solid
The company did not simply comply. It publicly dismantled the US justification - point by point.
First, on red-teaming. Before its launch, Fable 5 was tested for thousands of hours by specialist teams - internally, but also with the US government itself, the UK AISI and several independent third-party organisations. The result: no universal jailbreak was found. The model passed every test.
Next, on the nature of the flaw. The reported jailbreak is not a secret backdoor. The technique simply involves asking the model to read a codebase and identify its flaws - something thousands of developers do every day to protect their systems. This capability is already available in other public models such as OpenAI's GPT-5.5. Blocking Fable 5 therefore changes nothing in the balance of power.
Finally, on the dangerous precedent it sets. If the rule becomes "a model is withdrawn as soon as a minor flaw is detected", no AI will ever be released again. Every model has flaws - it is inevitable. The real question: was this flaw serious enough to justify a suspension affecting hundreds of millions of users? For Anthropic, the answer is clearly no.
The real issue: Europe has just been pushed out of the game
What happened with Fable 5 is not a technical incident. It is a major geopolitical signal - and French political reactions were immediate.
Benjamin Haddad, French Minister Delegate for Europe, said the decision marked "an accelerator of the geopolitical battle over AI" and that Europe could "not settle for being an open market dependent on technologies designed, funded and controlled elsewhere." Put simply: we let the Americans build the tools we depend on, and they have just reminded us that they remain in control of them.
Anne Le Hénaff, Minister for Digital Affairs, stressed that the major powers now regard strategic technologies as "instruments of power" - in the same way as energy or defence - and called for faster European investment in sovereign AI.
And Benjamin Haddad had no illusions: "Let's not fool ourselves: Democrats or Republicans, Donald Trump's successors will not reverse this kind of decision." US control over the most powerful models is a bipartisan state policy - like the export restrictions on chips that came before it.
The question raised by Fable 5 is simple and brutal: can you build an AI strategy on tools that a foreign government can switch off overnight?
- ✓ Claude Fable 5
- ✓ Claude Mythos 5
- ✓ Project Glasswing
- ✓ All other models
- ✕ Claude Fable 5
- ✕ Claude Mythos 5
- ✕ Project Glasswing
- ✓ Opus 4.8 / Sonnet / Haiku
Applies to all foreign nationals, including Anthropic's foreign employees based in the United States
The AI Sisters lesson
This affair confirms something we have been telling our clients from the start: mastering AI in the workplace is not about mastering a tool. It is about building a skill.
Claude remains our number 1 recommendation - and Opus 4.8 is here, excellent and available. But the teams we train also know how to use Gemini and Copilot. Not out of distrust of Claude - out of strategic intelligence.
An employee who understands how to prompt, how to structure a use case and how to evaluate an AI response will never be blocked by a decree from Washington.
That is exactly what we build with you.
The AI Sisters conclusion
Fable 5 was a genuine breakthrough. Its suspension within 72 hours is a rude awakening. AI is not only a question of performance - it is also a question of sovereignty.
In the meantime, Claude Opus 4.8 is here, it is excellent, and it is not at risk of being cut off tomorrow morning.
Want to train your teams on Claude - without depending on a single model?
Where does your company stand with AI? Take the free AI diagnostic in 5 minutes and get your action priorities.
.avif)
