Anthropic Blocks China's AI Access Loopholes

Anthropic is intensifying efforts to prevent Chinese companies from bypassing its strict AI usage policies. Firms such as Ant Financial have allegedly accessed Anthropic's AI tools, including Claude Code, through indirect routes like cloud services and foreign subsidiaries. Similarly, ByteDance reportedly offers a reimbursement scheme for engineers using personal Claude subscriptions via VPNs. These circumvention tactics, while not illegal, contravene Anthropic's terms of service, which expressly prohibit usage by Chinese entities.

Anthropic maintains one of the most stringent bans among US AI companies regarding access in China, implementing user verification and disallowing payments from Chinese banks. This contrasts with OpenAI, whose tools are more accessible to Chinese users via VPNs. The demand for US AI products, especially Anthropic's coding tools, highlights their value for training smaller models, a process known as "distillation."

The AI company is continuously updating its detection methods to identify and block unauthorized access, including targeting "transfer station" services that relay requests from mainland China. Despite these measures, enforcement remains challenging as Chinese users develop new evasion techniques. Anthropic explicitly prohibits facilitating access in unsupported regions like China and is committed to enforcing its policies with partner support.