On September 5, 2025, Anthropic, a U.S. - based AI company, made an immediate announcement that it would cease providing Claude services to any groups or their subsidiaries where Chinese capital holds a majority stake (over 50%). In essence, this decision implies that any enterprise, whether directly or indirectly controlled (with an ownership stake exceeding 50%) by Chinese entities, will be barred from utilizing Anthropic's services.

This restriction is not confined to mainland Chinese companies alone. It also encompasses their overseas subsidiaries, cloud - service intermediaries, as well as organizations that have investment entities backed by Chinese capital. Anthropic clarified that this measure is being implemented to mitigate legal, regulatory, and national - security risks.

On the very same day, Zhipu, a domestic large - scale model enterprise, unveiled a 'Special Relocation Plan for Claude API Users.' This plan is fully compatible with the Claude protocol. It enables users to make a seamless transition to the GLM model API by merely changing the API URL.