OKX Blocks Claude for Hong Kong Staff, Employees Traveling Through China

OKX has blocked use of Anthropic’s AI model Claude for employees based in Hong Kong and those traveling through China, underscoring how the US-China AI rivalry is reaching day-to-day corporate operations.

Bloomberg reported on Aug. 19 that OKX cut off Claude access for Hong Kong-based employees after its corporate account was temporarily suspended in early August. The corporate account was later restored.

In a statement, OKX said it recently learned that some Hong Kong-based employees’ use of Claude may not have fully complied with Anthropic’s regional access policy. The company said it would limit Claude access for Hong Kong employees when necessary and redirect requests to other suitable AI models. Anthropic told Bloomberg that its policy does not permit access to Claude in Hong Kong or China.

OKX has encouraged employees to make active use of AI tools. Staff have used both Claude and OpenAI’s ChatGPT for work, though the company has internally viewed Claude as the more effective tool. OKX is currently spending $6 million to $8 million a month on multiple large language model services. In an internal notice issued in early August, the company said it would not support Claude access via virtual private networks, or VPNs, for employees visiting Hong Kong or China. In a follow-up notice this week, it reiterated that Claude can be used only in “supported regions.”

OKX is not alone in restricting Claude. Bloomberg has also reported that Goldman Sachs blocked Hong Kong employees from accessing Anthropic’s models.

Separately, OKX sold part of its stake to Intercontinental Exchange Inc., the owner of the New York Stock Exchange, in March. The transaction valued the company at $25 billion.