Why Hong-Kong bankers of Goldman Sachs can't use Anthropic's models
What's the story
Goldman Sachs has barred its Hong Kong-based bankers from using Anthropic's artificial intelligence (AI) models. The decision comes as global financial institutions are becoming more cautious about AI tools due to increasing concerns over data security. Employees of the US bank in Hong Kong were previously able to use Anthropic's Claude via an internal AI platform. However, access has been revoked in recent weeks, a source familiar with the matter told Reuters.
Contractual concerns
Decision based on strict interpretation of contract with Anthropic
The decision by Goldman Sachs was taken after a strict interpretation of its contract with Anthropic. The bank consulted the company and concluded that its employees in Hong Kong shouldn't use any of Anthropic's products. This move is part of a broader trend among global banks and financial regulators, who are wary of potential risks posed by advanced AI models like Anthropic's latest offering, Mythos.
Access continuity
Other AI tools still accessible to Goldman Sachs employees
Despite the restriction on Anthropic's AI models, other mainstream tools like Gemini and ChatGPT are still available on Goldman Sachs's internal platform. The move to restrict access in Hong Kong comes amid rising tensions between the US and China over AI technology, data security, and access to advanced computing tools.
Market position
Speculation about other banks limiting access
Notably, Anthropic's API and Claude.ai are not officially available in Hong Kong, according to the company's disclosure. The move by Goldman Sachs has sparked speculation about whether other banks or companies will follow suit in limiting access to advanced AI models in the region.
Competition fears
Concerns over Chinese companies using American AI models
The restriction of Anthropic's AI model in Hong Kong by Goldman Sachs comes amid fears that Chinese companies are using American AI models to train their own systems at a lower cost. This practice, known as AI distillation, has been flagged by several US tech giants. In February, OpenAI told US lawmakers it had caught DeepSeek trying to secretly copy its most powerful AI models, and warned that the firm was developing new methods to disguise its actions.