Goldman Sachs blocks Anthropic Claude access for bankers in Hong Kong

In short: Goldman Sachs has stopped its bankers in Hong Kong from using Anthropic’s Claude AI models.

What happened

Goldman Sachs employees in Hong Kong have been unable to access Anthropic’s Claude models for the past few weeks, according to four people familiar with the situation.

The restriction applies even when staff try to use Claude through Goldman’s internal AI tools. A person familiar with the move said Goldman made the change after talking with Anthropic and then taking a strict reading of its contract with the start-up.

Anthropic said its Claude models had never been officially “supported” in Hong Kong, but it did not explain further. Goldman Sachs declined to comment.

This comes amid wider US and China tensions around AI. Western AI chat tools like OpenAI’s ChatGPT and Anthropic’s Claude are blocked in mainland China. Hong Kong has usually had more open access, but the Financial Times reports that limits can also be set by the US AI companies themselves.

One concern mentioned in the report is “distillation”, which is when someone uses a powerful AI system heavily to help train a new, competing system (like learning a recipe by tasting a dish many times). US officials have also accused China of large-scale theft of AI-related intellectual property, while China has rejected those accusations.

Why it matters

Many finance workers use AI tools to help write computer code and build financial models. If some teams cannot access the same tools as colleagues in other places, it could slow work down or create uneven playing fields across offices. The change could also raise questions for other multinational firms in Hong Kong that have company-wide deals with Anthropic.

Source: Financial Times