Why Anthropic Is Banning Claude Access for Companies Linked to China and Other ‘Adversary’ Nations

On September 5, Anthropic, the developer of the large language model Claude, announced an update to its sales restrictions, immediately halting Claude usage for groups or subsidiaries of companies in which Chinese capital holds a majority stake.

The policy, the first of its kind by a U.S. AI firm, applies to any company directly or indirectly owned more than 50% by entities from unsupported jurisdictions, regardless of where the company operates.

Anthropic cited legal, regulatory and security risks as the rationale, warning that such entities could be compelled to share data with intelligence agencies or develop AI capabilities that might be used against U.S. and allied interests.

The restriction also extends to countries the United States deems “adversary states,” including Russia, Iran and North Korea, affecting direct customers and groups accessing Anthropic services via cloud platforms.

According to a senior executive, the change could impact Anthropic’s global revenue by several hundred million dollars.

Anthropic’s full statement emphasizes strengthening regional limits to align with its commitment to democratic values and to mitigate national‑security risks associated with transformative AI.