How people in China continue to outsmart humanity’s geolocation limitations

Anthropy goes to China has made great efforts to prevent people in China from using AI models, but in practice, its safeguards have often failed. Over the past year, startups, researchers, and technology enthusiasts across the country have developed increasingly sophisticated solutions for accessing Cloud. Many of them consider it the most capable AI assistant in the world, which makes the extra effort to get it worthwhile.

Anthropic goes to great lengths to prevent people in China from using its AI models, but in practice, its safeguards often fail. Over the past year, startups, researchers, and technology enthusiasts across the country have developed increasingly sophisticated solutions for accessing Cloud. Many of them consider it the most capable AI assistant in the world, which makes the extra effort to get it worthwhile.

In early June, Anthropic publicly released Fable 5, a protected version of its most powerful AI model to date, Mythos. Chinese social media immediately lit up with posts from people who shared their impressions after trying it out. (Anthropic Access has been revoked to the global model a few days later in response to export controls imposed by the Trump administration).

Chinese people generally have access to other Western AI tools, such as OpenAI’s ChatGPT, using VPNs, foreign phone numbers, and international payment methods to create and maintain their accounts. But Anthropic has arguably taken more aggressive steps, such as banning accounts it suspects are owned and controlled by people based in China. On Chinese social media, users often report that their Cloud account was suspended without warning, despite taking those precautions.

The cat-and-mouse game has fueled a thriving underground economy to gain access to Claude in China. The accounts are being sold on Chinese e-commerce platforms such as Taobao and through illicit marketplaces on Telegram. More recently, a cottage industry of “transfer stations” has also emerged. These services act as intermediaries, purchasing access to Anthropic’s API outside China and then redistributing Claude API tokens to users within the country. This setup is designed to give startups and other professional users more stable and reliable access to the AI ​​assistant.

Michael Aciman, a spokesman for Anthropic, says the company uses a range of sophisticated detection systems, including identity verification, to enforce its policies against unauthorized access to the Cloud. He added that Anthropic also worked to detect and disrupt proxy networks used to provide chatbot access in China.

Despite all the difficulties that the Chinese people had to overcome to use Claude, there are still many loyal fans of Anthropy in the country. It is especially popular among programmers. Although Chinese companies like DeepSeek and Z.ai have some of the most capable open source large language models on the market, third-party tests still show that they lag behind leading closed models like Cloud. During a recent reporting trip to China, WIRED spoke to academics and engineers at several technology companies who said they prefer using Claude over Chinese models for creating code, and are eager to try every new model Anthropic releases.

The research was conducted by Zilan Qian, a research associate at the Oxford China Policy Lab Black market To resell Western AI tokens to Chinese users. He noted that Chinese software developers say they overwhelmingly prefer to use tools like Claude Code and OpenAI’s Codex compared to those of local companies. “The analysis shows that the Chinese models are still six to nine months behind the US models, and for specific things like programming and development, you can clearly see the gap,” Qian says.

“For Chinese AI policymakers and technologists, they have much less problem relying on and using American ideas or products, regardless of geopolitical or ideological rivalry,” says Matt Sheehan, a senior fellow at the Carnegie Endowment for International Peace, who researches AI policy and China. “It’s Americans who tend to think that an idea or product is tainted simply because it comes from a competitor,” he says.