On September 5, 2025, Anthropic announced that it would immediately stop providing Claude to groups or their subsidiaries whose majority shares are held by Chinese capital. This move means that any company that is directly or indirectly controlled by a Chinese entity (more than 50% of the shares) is no longer allowed to use Anthropic’s services.
This policy applies not only to mainland Chinese companies, but also to organizations that have overseas subsidiaries, cloud service transit entities, or investment entities with Chinese backgrounds.
Anthropic stated in an announcement on its official website that this move was in response to legal, regulatory and national security risks.
In addition to Chinese companies, the policy also applies to other entities regarded as “adversarial nations” by the United States, such as Russia, Iran, North Korea and other entities.
The move is also intended to curb the possibility of Chinese companies circumventing export controls to acquire advanced AI technology by registering subsidiaries overseas, such as in Singapore, or using third-party cloud services, according to a briefing an Anthropic executive told the Financial Times.
This is the first case in which a U.S. AI company has publicly announced such restrictions, signaling that U.S. technology companies may be more proactive in taking preventive measures regarding restrictions on AI exports and services.