The world of artificial intelligence in 2026 finds itself at a point where technological progress is colliding with the harsh reality of national borders and trade restrictions. While companies like Anthropic strive to provide the most advanced LLMs (Large Language Models) on the market, their availability is increasingly dictated by geopolitical pressures. The recent decision by JPMorgan Chase (JPMC) to block Claude models for its Hong Kong branches is a clear signal that the era of uncontrolled global AI proliferation is coming to an end.
The Licensing Trap: Why Did JP Morgan Say "No" to Anthropic?
According to Financial Times reports, the reason is not the unreliability of Claude models. The problem lies in the actual licensing terms that Anthropic sets for its commercial partners. Specifically, it concerns an explicit ban on using the models in the Greater China region, which also includes Hong Kong and Macau.
For global corporations like JPMorgan, these terms represent significant operational risk. If a company uses AI to analyze sensitive financial data, it must have absolute certainty that the software it uses is compliant with both local and international laws. In this case, it turned out that the strict interpretation of Anthropic's terms creates an impassable barrier for the banking segment in Asia. A similar step was already taken in April by Goldman Sachs, indicating a broader trend of major financial institutions moving away from Anthropic in favor of other providers in certain regions.
Comparison: Claude vs. the Competition
To understand the situation, it is important to know what Anthropic is up against. The Claude 3.5 series models (and their subsequent versions in 2026) are considered top-tier in many aspects, especially in logical reasoning and natural language fluency. In benchmark tests, they often outperform GPT-4o from OpenAI and models from Google (Gemini) in tasks requiring a high degree of ethical reasoning and precision in code writing.
- Claude (Anthropic): Excels in creative writing and complex logic. Priced affordably through Claude Pro at around 20 USD/month (approx. 470 CZK) or via API based on usage.
- GPT-4o (OpenAI): Multimodal leader with the broadest application ecosystem. Price similar to Claude.
- Gemini (Google): Strong integration player, especially thanks to Google Workspace connectivity.
The Anthropic Paradox: AI at Work vs. AI in Hiring
An interesting and contradictory aspect of Anthropic's operations is their approach to human resources. While their products are designed to maximize efficiency, during their own employee hiring process the company has issued an unusual stance: they ask applicants not to use AI tools when submitting their applications.
The reason is not an attempt to slow down work, but the need to verify a person's actual communication skills. Anthropic wants to see authentic interest and the ability to formulate thoughts without LLM assistance. As researcher Simon Willison notes, this strategy emphasizes the importance of building a trustworthy environment. If an employee is unable to communicate without AI, it could in the future lead to masking errors or unethical use of tools when solving critical tasks.
Impact on the Czech Market and the European Scene
What does this event mean for us in the Czech Republic? Even though Hong Kong may seem distant, the situation has direct implications for European companies and developers:
- Availability in the Czech Republic: Claude models are fully available in the Czech Republic and work very well even with Czech localization. However, Czech companies must monitor whether similar licensing restrictions will spread through European regulations as well.
- EU AI Act: The European Union places great emphasis on transparency and security. If companies like Anthropic become involved in legal disputes (e.g., with the US Pentagon over blacklisting), it could affect their ability to meet the strict rules of the EU AI Act, which could lead to restrictions on their services in our market as well.
- Data Sovereignty: For Czech companies, it is crucial to know where data is processed. Geopolitical conflicts can cause a tool you use today to be suddenly banned due to a political decision within corporate policy.
Practical impact for you: If you are a freelancer or work in a company, do not lock yourself into a single model. Diversification (using a combination of GPT, Claude, and possibly open-source models like Llama) is the best insurance against geopolitical upheavals that can change the availability of your favorite tools overnight.
Is Claude fully functional in Czech?
Yes, Anthropic models show a very high level of Czech language comprehension and can produce text that sounds natural, which is one of their main competitive advantages over some other models.
Why would a company ban AI that increases productivity?
In JPMorgan's case, it is about minimizing legal risk. If a licensing agreement prohibits the use of a model in certain regions (such as China/Hong Kong), the company must block access to avoid contract violations and subsequent massive fines.
Could this affect me in the Czech Republic?
Not directly, but indirectly yes. If global AI providers come into conflict with regulators or political blocs, services could be restricted within the EU as well if those services do not fully comply with European security and data sovereignty standards.
