Goldman has restricted its Hong Kong staff from using Anthropic’s Claude AI model, with the restriction reportedly rooted in geographic considerations.

 

OKX’s restriction on its staff using Claude AI came shortly after the exchange’s Claude account was suspended by Anthropic over noncompliance with its geographic access policies.

 

Although the suspension of the exchange’s Claude account lasted only a short period, an internal staff message said that OKX would not support the use of Virtual Private Network (VPN) access to Claude for employees visiting Hong Kong or China.

 

As part of its plans to continue allowing its Hong Kong staff to use AI in their daily workflows, OKX said it would direct its Hong Kong employees to other supported AI models. Meanwhile, Star Xu, OKX CEO, has said that the exchange spends between $6 million and $8 million on multiple large language model (LLM) providers and has even built Oli, an internal AI development platform.

 

OKX’s restriction on its Hong Kong staff comes shortly after Goldman Sachs imposed a similar restriction on its Hong Kong staff earlier this year. However, unlike OKX’s restriction, which was driven by an account suspension, Goldman Sachs’ restriction stemmed from a strict interpretation of its contract with Anthropic.

 

After reviewing its contract with Anthropic, Goldman Sachs concluded that the agreement did not permit the use of Anthropic or any of its products, including Claude AI, in Hong Kong, and thus restricted their use.

 

However, this did not mean that Goldman Sachs was restricting the use of Anthropic’s products or Claude AI overall. In fact, Goldman Sachs has partnered with Anthropic engineers to build AI agents for its trade accounting and client vetting.