JPMorgan Chase has stopped employees in Hong Kong from accessing Anthropic’s artificial intelligence models through the bank’s internal systems. The move follows a similar restriction introduced by Goldman Sachs for its Hong Kong bankers in April.
The Financial Times reported the decision on Thursday, citing three people familiar with the matter. JPMorgan and Anthropic did not respond to requests for comment outside business hours. Therefore, the companies have not publicly confirmed the details of the restriction.
JPMorgan employees in Hong Kong can no longer select Anthropic’s Claude models from an internal list of approved large language models. Staff previously accessed the technology through the bank’s controlled AI platform rather than through public consumer services.
According to the report, JPMorgan acted after reviewing the wording of Anthropic’s usage terms under its licensing agreement. The restriction appears to apply to Claude models offered through the internal platform. The report did not state how many workers lost access or whether JPMorgan plans to restore the service.
The reported change places JPMorgan alongside Goldman Sachs, which removed Claude access for its Hong Kong bankers in April. Goldman employees had also used Anthropic models through an internal system before the bank withdrew the option.
Goldman Sachs reportedly made its decision after reviewing its agreement with Anthropic. Yet neither bank has said that employees misused the models or exposed company data. The available reports link both actions to contract terms and geographic access rules.
Anthropic does not list Hong Kong among the locations where Claude.ai and its application programming interface are officially supported. Mainland China is also outside the company’s supported regions. Still, some global businesses have provided access through company-wide contracts and systems hosted in other locations.
That arrangement allowed some employees in Hong Kong to use US-developed AI models through approved workplace platforms. Nevertheless, the JPMorgan and Goldman restrictions show that global contracts may not provide the same access in every market.
The decisions come as US technology companies face closer checks over advanced AI services offered outside the country. US and Chinese authorities have also increased controls covering advanced chips, computing systems, data and artificial intelligence tools.
Hong Kong operates under a separate legal and internet system from mainland China. Even so, US companies can set their own service limits for the city. Contract terms, export controls and internal compliance policies can also shape whether workers receive access.
Earlier in June, the US government ordered Anthropic to suspend foreign-national access to Fable 5 and Mythos 5. The directive applied to foreign nationals inside and outside the United States, including Anthropic employees who are not US nationals.
Anthropic said the order forced it to disable both models for all customers while it worked to meet the government’s requirements. The company stated that access to its other Claude models was not affected. It also said authorities had not provided details about the national security concern.
“The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance,” Anthropic said. The company added that it believed the directive involved a ‘misunderstanding’ and said it was seeking to restore access.
US officials reportedly raised concerns that foreign military or intelligence users could misuse advanced AI tools. Still, authorities did not publicly identify a customer or incident linked to the models. The JPMorgan restriction was reportedly based on licensing language, and no public evidence shows that the bank acted directly under the government order.
Also Read: Anthropic Faces White House Heat as AI Curbs Threaten IPO Plans