JP Morgan Chase (JPMC) has blocked its Hong Kong staff from accessing AI models developed by Anthropic, according to the Financial Times.

Citing three people with close knowledge of the decision, the paper said that JPMC employees in the region are no longer able to select Claude models from an internal list of approved large language models (LLMs).

Another source told the FT that JPMC blocked Anthropic’s models due to the wording of the usage terms in the pair’s licensing agreement.

US AI models are not available in China, but to date international firms based in Hong Kong have been able to access them via internal corporate networks on servers hosted outside the region.

JPMC declined a request for comment.

The news follows the FT’s April report that Goldman Sachs had removed access to Claude for Hong Kong staff, with sources saying the bank had taken the move due to a strict interpretation of Anthropic’s terms of use.

Another source said the specific issue lies in Anthropic’s explicit ban on using its models in Greater China, including territories such as Hong Kong and Macau.

Anthropic models have been subject to repeated bans throughout 2026. In March, the firm sued the Pentagon for blacklisting its models after the Trump administration designated it a “supply chain risk,” a decision that Anthropic executives have said could cost billions in lost revenue.

On 10 June Microsoft blocked its employees from using Fable 5, Anthropic’s new frontier LLM, over data retention fears. The AI company disabled the model for all users on 12 June following a US government order to prevent foreign nationals from accessing both Fable 5 and Mythos 5.

---