Fable 5 disabled: why Anthropic’s US block matters for Europe and AI sovereignty

On Friday, June 12, 2026, the US government ordered Anthropic to restrict access to its two most capable models, Claude Fable 5 and Claude Mythos 5, for any foreign national. The order applied inside and outside the United States, including to non-US Anthropic employees. Anthropic said it could not verify every user’s nationality in real time, so it disabled Fable 5 and Mythos 5 for all users rather than trying to make a live eligibility decision on each request. Other Claude models were not affected.

The live status widget above this article checks whether Anthropic Fable 5 is currently being served to our European Anthropic account. It is not a general Claude status page, and it does not replace Anthropic’s own communications. It is a practical signal for European users asking a simple question: is Fable 5 actually available from here right now?

On Friday, June 12, 2026, Anthropic received a US export-control directive requiring it to suspend access to Fable 5 and Mythos 5 for foreign nationals, whether located inside or outside the United States.
Because Anthropic could not segment access by nationality in real time, it disabled both models for all users. Other Claude models remained available.
The stated rationale was national security, tied to an alleged jailbreak method. Anthropic disputed the characterization, described the issue as narrow, and criticized the lack of a transparent process.
The important point is not only that Fable 5 was blocked. It is the precedent: access to frontier AI can become dependent on nationality, jurisdiction, and geopolitics.
For Europe, the incident turns AI sovereignty from policy language into an operational continuity question.

What happened

Anthropic announced that it had received a US government export-control directive requiring the company to suspend access to Claude Fable 5 and Claude Mythos 5 for foreign nationals.

According to Anthropic, the government cited national security authorities. Anthropic’s understanding was that the concern related to an alleged method for bypassing, or “jailbreaking”, Fable 5. Anthropic disputed the characterization of the risk, saying the demonstrated capability appeared narrow and comparable to capabilities already available in other deployed models.

The immediate result was clear: Anthropic disabled access to Fable 5 and Mythos 5 for all customers. Access to other Anthropic models was not affected.

This is the core fact behind the surge in searches asking whether Fable 5 was disabled or blocked, and why Anthropic suspended Mythos 5 alongside it. The interruption was not a normal outage; it was a compliance response to a government directive.

What the US directive ordered

The directive required Anthropic to suspend access to Fable 5 and Mythos 5 by foreign nationals, whether those people were located inside or outside the United States.

That distinction matters. The rule was not simply geographic. It was not only “outside the US”. It covered foreign persons, including non-US Anthropic employees.

For a commercial AI platform, that creates a hard operational problem. Most AI providers can determine billing region, IP region, account metadata, enterprise contract status, and sometimes residency-related details. They generally do not have a reliable real-time mechanism for determining every user’s nationality at the moment a model request is made.

Anthropic’s conclusion was that the only way to ensure compliance was to disable access broadly.

Why the precedent matters

Models rise and fall constantly, and that churn is normal. A frontier capability vanishing overnight by government order is not.

Frontier AI models are becoming part of everyday work: software development, research, legal analysis, customer support, internal operations, data analysis, and security work. If access to the strongest models can change suddenly because of national-security decisions in another jurisdiction, companies outside that jurisdiction inherit a new kind of operational risk.

This does not mean governments should have no role in AI safety. The issue is process, scope, and predictability.

A transparent safety process is one thing; abruptly cutting a commercial model off from users worldwide is another. For European companies, the takeaway has nothing to do with US providers being unreliable. It is simpler: leaning on a single provider, country, or legal regime is fragile by design.

What it means for Europe

For Europe, the Fable 5 and Mythos 5 case turns AI sovereignty from an abstract policy phrase into an operational question.

A company may be European, serve European customers, comply with GDPR, and still depend on AI infrastructure controlled by providers and governments outside the EU. That dependence can be manageable, but it should be visible. If it is invisible, it becomes a single point of failure.

“Claude blocked in Europe” is not the full story here, because other Claude models remained available and the restriction was tied to Fable 5 and Mythos 5. But the concern behind that search is real: European users need to know whether critical AI capabilities can be interrupted by decisions made outside Europe.

AI sovereignty in Europe does not mean rejecting every non-European model. It means keeping strategic choice: knowing where data is processed, which providers are involved, which legal regimes apply, and what fallback paths exist if a model is disabled, degraded, repriced, or removed.

Does Europe need a European alternative to ChatGPT, Claude, and Gemini?

The demand for European AI is not only about branding. It comes from concrete requirements: regulated companies need clearer control over data handling, public-sector teams need procurement and sovereignty guarantees, legal and finance teams need predictable governance, and enterprises need continuity plans when providers change access rules.

That is why the search for a European alternative to ChatGPT, Claude, and Gemini should not be reduced to a single replacement product. For companies, the better question is usually architectural: how do we avoid building knowledge work, customer workflows, and internal automations on one model provider with no exit path?

A resilient AI stack can include US models, European models, open models, private deployments, and hosted regional options. The goal here is operational leverage, not ideological purity.

GDPR, ChatGPT, Claude, and sovereign AI

GDPR and AI sovereignty are related, but they are not the same thing.

GDPR asks whether personal data is processed lawfully, transparently, securely, and with appropriate rights and safeguards. A company using ChatGPT, Claude, Gemini, or another AI system still needs to understand what data is sent, where it is processed, which subprocessors are involved, how retention works, and what contractual protections apply.

AI sovereignty asks an additional question: who controls access to the capability itself?

The Fable 5 disabled incident shows why both questions matter. Even if a company has a compliant data-processing arrangement, access to a frontier model can still be shaped by export controls, national-security policy, provider decisions, and jurisdictional constraints.

For European organizations, GDPR compliance is necessary. It is not a complete resilience strategy.

ilisai’s multiprovider approach

ilisai is not a frontier model lab, and it should not be described as one. It is a multiprovider AI platform designed to reduce dependence on a single model provider.

That distinction matters. No single platform can make geopolitics disappear; the practical response to Fable 5 being blocked is to build AI workflows that keep provider choice open.

A multiprovider approach helps teams compare models for each use case, move workloads when a model is unavailable, support regional access options where providers make them available, and evaluate European AI options alongside global frontier models.

For European companies, that is the pragmatic path: use the best available models, but do not let one provider become the only path through which work can happen.

All your AI tools in one place

Access GPT, Claude, Gemini, and more — compare models and find what works best for you.

Explore features

FAQ

Is Claude a European AI?

No. Claude is developed by Anthropic, a US company. European organizations can use Claude where it is available and where their legal, security, and data-protection requirements are satisfied, but Claude itself is not a European AI system.

What happened to Fable 5?

Anthropic disabled Claude Fable 5 after receiving a US government directive restricting access to Fable 5 and Mythos 5 for foreign nationals. Anthropic said it could not verify every user’s nationality in real time, so it removed access broadly to comply.

Is Fable 5 available in Europe?

Fable 5 was disabled as part of Anthropic’s response to the US directive. The live widget above this article checks whether Fable 5 is currently being served to our European Anthropic account. Other Claude models were not affected by the same directive.

What European alternatives to ChatGPT exist for companies?

Companies can look at European model providers, open models deployed in controlled environments, EU-hosted AI infrastructure, and multiprovider platforms that let teams choose between models instead of relying on a single vendor. The right alternative depends on the use case, data sensitivity, required quality, hosting region, procurement needs, and compliance requirements.

What does it mean to use AI platforms hosted in the EU?

Using AI platforms hosted in the EU generally means that infrastructure, processing, or deployment options are located within the EU. That can help with data-residency, procurement, latency, and governance requirements, but it is not enough on its own. Companies still need to review the provider, subprocessors, retention policies, security controls, model origin, and applicable legal terms.

What do GDPR, Anthropic, and ChatGPT have to do with this?

GDPR governs how personal data is handled. Anthropic and ChatGPT are examples of major AI providers that companies may use to process work-related or personal data. The Fable 5 incident adds another layer: even when data protection is addressed, access to AI capabilities can still depend on provider policy, export controls, and jurisdiction. European companies need to evaluate both compliance and resilience.