Anthropic, the San Francisco AI company behind the Claude chatbot, has moved to block Chinese-owned companies -- including subsidiaries abroad -- from using its technology.
The firm said the decision is aimed at protecting U.S. national security and closing loopholes that allowed restricted actors to access its models through overseas fronts.
Even if a subsidiary operates in Europe or North America, Anthropic argues it remains bound by Chinese law and vulnerable to government pressure. That link, the company warned, could funnel American AI advances into surveillance systems, censorship tools, or autonomous military projects.
China now joins Russia, Iran, and North Korea on Anthropic's blacklist. Executives said authoritarian states had exploited foreign registrations in the past to bypass restrictions, a practice the company is determined to end.
Anthropic's move marks a break from the tech industry's traditional reluctance to engage in foreign policy. Chief executive Dario Amodei has long pushed for stricter export controls, warning that rivals such as Alibaba, Tencent, ByteDance, and especially DeepSeek are catching up fast. U.S. regulators have already banned federal use of DeepSeek, highlighting the security risks posed by advanced Chinese AI platforms.
The restrictions could cost Anthropic hundreds of millions in potential revenue, but the company appears willing to absorb the loss. Backed by Amazon and valued near $183 billion, Anthropic is rapidly growing, with more than 300,000 enterprise customers and a rising number of six-figure accounts. Leaders say the long-term benefits of protecting U.S. interests outweigh the short-term financial downside.
By prioritizing safety over profit, Anthropic is signaling that advanced AI firms see themselves as part of national defense infrastructure. The company believes restricting access to authoritarian states not only safeguards its own technology but also helps maintain America's edge in the global AI race.