Claude Completely Banned for Chinese Companies – No Workarounds Anywhere

On 5 September 2025 Anthropic announced a policy that immediately bans the use of Claude for any entity controlled by China.

The ban covers:

Companies incorporated in mainland China;

Overseas subsidiaries in which Chinese shareholders own more than 50 %;

Entities that access Claude indirectly through cloud services, third‑party platforms, or investment structures with a Chinese majority.

Anthropic cited legal, regulatory and security risks as the rationale. The company expressed concern that unrestricted access could enable the development of military or intelligence applications and that techniques such as model distillation might allow Chinese firms to capture Claude’s capabilities, accelerating their own large‑model research and creating direct competition with U.S. and allied technology firms.

Choosing to forgo potential revenue, Anthropic closed the service to the listed entities and simultaneously urged the U.S. government to strengthen export controls, accelerate domestic energy and compute infrastructure, and rigorously assess AI model safety to prevent frontier technologies from being exploited by adversary nations.

In practice, the restriction also intercepts API calls made via cloud platforms. Even subsidiaries registered in Singapore, the United States, or other jurisdictions are blocked if the controlling ownership is predominantly Chinese.

For multinational teams that rely on Claude’s API, the sudden shutdown can halt ongoing projects that depend on the service.

Reference URLs:

https://www.anthropic.com/news/updating-restrictions-of-sales-to-unsupported-region

https://www.ft.com/content/12b8e10b-b55d-4824-817f-a3c9cfe9f779