JPMorgan Blocks Anthropic’s Claude for Staff in Hong Kong: Report

JPMorgan Chase has blocked employees in Hong Kong from accessing Anthropic’s Claude models, according to a Financial Times report, making it the latest Wall Street bank to tighten controls on the use of advanced AI models in the Asian financial hub.

The move affects Claude models available through the bank’s internal platform, where employees can select from a list of approved large language models (LLMs). The report added that JPMorgan’s decision was linked to the wording of Anthropic’s licensing agreement and usage terms.

The restriction comes months after Goldman Sachs adopted a similar policy for its Hong Kong operations. According to the Financial Times, Goldman interpreted Anthropic’s terms of service as prohibiting the use of Claude models across Greater China, including Hong Kong.

The development highlights growing tensions over access to frontier AI models outside the United States, particularly as governments and companies become increasingly concerned about the strategic and security implications of advanced AI models.

Last week, Anthropic suspended access to Claude Fable 5 and Mythos 5 after receiving a US government directive requiring the company to block foreign nationals from using the systems. The order, issued under export control and national security authorities, applied to all foreign citizens, whether inside or outside the United States, including Anthropic’s own foreign-national employees. 

While services such as OpenAI’s ChatGPT and Anthropic's Claude are inaccessible in mainland China due to Beijing’s internet controls, Hong Kong has traditionally remained outside those restrictions. 

International firms operating in Hong Kong have historically worked around geographic restrictions through global enterprise agreements and by hosting AI workloads outside China. The latest actions by major banks could complicate the use of cutting-edge AI tools in the territory, where financial institutions have been rapidly adopting generative AI for tasks such as software development and improving employee productivity.

US AI companies have also become more cautious about the risk that their models are being used to support the development of competing systems in China. One concern is distillation, a process through which developers can use outputs from advanced models to train their own AI systems.