Crypto exchange OKX has restricted employees in Hong Kong from using Anthropic’s Claude after the company’s corporate account was briefly suspended.
US-China tensions over artificial intelligence are affecting global businesses, and this is the latest example of this.
Geographic access rules
OKX stopped providing Claude access to employees in Hong Kong and to staff traveling through China in early August, according to Bloomberg. The company’s corporate account has since been restored.
The employees’ AI requests are being processed by other models.
Anthropic’s policies prohibit access to Claude from unsupported locations. Its current list of supported locations does not include Hong Kong or mainland China.
AI as part of daily work
The restrictions are particularly notable for OKX because the exchange has been pushing employees to use AI.
In fact, AI proficiency has become part of employee performance evaluations.
OKX said it spends between $6 million and $8 million per month across several major large-language-model providers.
Employees had previously been able to use both Claude and OpenAI’s ChatGPT.
Other similar cases
This is not an isolated incident. Goldman Sachs faced a similar situation earlier this year. In April, the bank’s employees in Hong Kong lost access to Claude, with the restriction applying even to staff from other countries. Claude was used mainly by Goldman’s software engineers.
Chinese technology giant Alibaba took the opposite approach in July. It actually banned its own employees from using Claude Code. The company classified Anthropic’s coding assistant as high-risk software.
That policy has become increasingly difficult for companies to circumvent. The Financial Times reported in July that firms including Ant Financial and ByteDance had sought access through overseas subsidiaries, but new restrictions make this challenging.
