Anthropic's efforts to prevent Chinese users from accessing its AI models have been consistently thwarted by determined individuals and startups. Despite the company's aggressive measures, including account bans and identity verification, users in China continue to find creative workarounds to access Claude, Anthropic's AI assistant. The demand for Claude is particularly high among programmers, who prefer its capabilities over domestic alternatives. To meet this demand, a thriving underground economy has emerged, with accounts and API tokens being sold on e-commerce platforms and illicit marketplaces. Recently, 'transfer stations' have become a popular solution, acting as intermediaries between Chinese users and Anthropic's API. These services purchase access to the API outside of China and redistribute tokens to users within the country, providing a more stable and reliable access point. However, this cat-and-mouse game raises concerns about security risks and the potential for malicious actors to exploit these unsanctioned tools. As Anthropic continues to tighten its restrictions, it is likely that Chinese users will continue to find ways to access Claude, highlighting the need for AI safety experts to consider the implications of this underground infrastructure.

Source: Wired