Today, June 13, 2026, will be remembered by people in the technology world as the moment when the boundaries between technological progress and state security definitively collapsed. What was supposed to be a celebration of a new generation of intelligence has become the subject of political coercion. Anthropic, known for its emphasis on safety (so-called AI Safety), finds itself in a situation where it must prioritize government regulations over technological development.
Finding from the perspective: Why Mythos 5 and Fable 5 specifically?
The reason why US Secretary of Commerce Howard Lutnick intervened now is due to the capabilities of the new models. According to NBC News reports, the main concern is the ability of the Mythos 5 model to perform "vulnerability chaining".
In technical practice, this means that the model can not only find a bug in software code but can intelligently link these errors to create a complex attack that allows it to take control of systems. This capability is incredibly useful for a regular programmer when searching for bugs, but for a cybercriminal, it represents a weapon of mass destruction. The Fable 5 model, which is essentially a modified and "trimmed-down" version of Mythos, was launched just days before the restrictive measures were adopted.
Performance Comparison: How powerful are these models?
To understand the scope of the problem, we need to look at how these new models stand in the context of current competition. While models like GPT-5 from OpenAI or Gemini 2.0 from Google focus on broad versatility and multimodality, Anthropic's Mythos series was designed with an extreme emphasis on deep logical reasoning and code analysis.
| Model | Main Focus | Security Profile |
|---|---|---|
| Mythos 5 | Advanced code, logic | Extremely high risk (government stopped) |
| GPT-5 | Multimodality, creativity | High (strict filters) |
| Gemini 2.0 | Google Ecosystem, data | Medium to high |
Impact on users in the Czech Republic and the EU
What does this mean for us? For a Czech user who uses Claude (e.g., via a web interface or API) for data analysis or programming, the situation is complicated. Anthropic stated in an official statement that it cannot effectively sort users by nationality without the risk of error. This means that access to Fable 5 and Mythos 5 models was likely immediately cut off for all users outside the USA, including us in the Czech Republic.
If you subscribed to Claude Pro (approx. 20 USD / month, which is roughly 470 CZK) expecting to have access to the latest technology, you will likely revert to older, but still very capable, Claude 3.5 series models. For Czech companies that build their services on the Anthropic API, this is a critical problem in planning capacity and service stability.
From the perspective of European regulation (EU AI Act), this situation is very interesting. While the EU strives for safe and transparent AI deployment, the USA is now choosing the path of "technological nationalism". This step could lead to market fragmentation – developers in Europe may be forced to use other models (e.g., Llama from Meta or local European projects) to avoid geopolitical fluctuations.
Prices and Availability: What remains?
The good news, then, is that Anthropic's standard models remain available. If you use Claude for general text generation, document analysis, or assistance with Czech (which Claude handles at a top-notch level), your work will not be interrupted.
- Claude Free Tier: Still available, but with a limited number of messages.
- Claude Pro: Approx. 20 USD/month – still available, but without Fable and Mythos.
- Claude Team: Approx. 30 USD/user/month – for companies with higher limits.
Editor's Conclusion: The end of uncontrolled progress?
This incident shows that the "wild west" era in AI is ending. Once models reach a certain level of capability (in this case, the ability to exploit software), they become a matter of national security, just like nuclear technology or advanced weapons. For us in the Czech Republic, this means we must be prepared for the most powerful tools to be regulated much more strictly than we had imagined.
Will I be able to use Mythos 5 models in the Czech Republic if I have US citizenship?
According to Anthropic's statement, the ban is aimed at foreigners. If you can prove your status as a US citizen, access should theoretically be possible, but Anthropic warns that the verification process is complex and errors may occur during account deactivation.
Will this decision affect Czech language support in Claude models?
No, this decision only concerns specific model architectures (Fable 5 and Mythos 5). The language capabilities of existing Claude models remain unchanged, and Czech is still fully supported.
Is this in line with the EU AI Act?
The EU AI Act focuses on risks associated with citizens' rights and transparency. The American decision is motivated by national security (geopolitics), which is a different framework than European regulation, but both directions aim for greater control over "high-risk" AI systems.
