Update to Sales Restriction Rules for Unsupported Regions

Anthropic prohibits the use of its services in certain regions due to legal, regulatory, and security risks. However, the problem of companies from adversarial nations such as China circumventing these restrictions by using overseas-established subsidiaries to access the service has persisted.

Companies controlled by authoritarian states may be compelled to share data or cooperate with intelligence agencies, which poses serious national security risks. This is because they could exploit Anthropic's models to develop applications for hostile military or intelligence services, or misuse model distillation techniques to boost their home country's AI capabilities.

Accordingly, Anthropic is strengthening its regional restriction rules as follows.

• Regardless of the region of operation, companies that are directly or indirectly 50% or more owned by a company headquartered in an unsupported region will be prohibited from using the service.

Additionally, Anthropic supports strong export control policies to prevent the leakage of technology that threatens national security, and continues to call for expanded AI infrastructure within the US and stronger security evaluations of models.