JPMorgan Chase cuts off Anthropic access for its Hong Kong staff

JPMorgan Chase has stopped employees in Hong Kong from accessing Anthropic’s Claude models, according to the Financial Times, in a move that underscores how AI availability is being narrowed across Greater China.

The report said Claude was removed from JPMorgan’s internal approved-model list for Hong Kong staff. People familiar with the matter told the FT that the decision was tied to Anthropic’s usage terms in the bank’s licensing agreement. JPMorgan declined to comment, and Anthropic did not immediately respond.

The restriction follows a similar action by Goldman Sachs, which also limited Anthropic access for staff in Hong Kong. The pattern suggests that major Wall Street banks are adjusting internal AI rules as they navigate regional licensing terms and broader U.S. export-control pressure.

Why Hong Kong matters

Hong Kong has historically had broader access to Western internet and technology services than mainland China, making bank-level restrictions there notable. Anthropic’s terms reportedly exclude use in Greater China, which includes Hong Kong.

That creates a practical issue for global banks with staff in the city: tools that may be available in the United States can still be restricted locally, even when the company has a broader enterprise relationship in place.

Policy pressure around AI access

The latest move comes after The Wall Street Journal reported on June 13 that Anthropic disabled its top AI models worldwide after a U.S. government export-control directive barred access by foreign nationals outside the United States.

Earlier reporting also showed Goldman Sachs chief executive David Solomon saying the bank was “hyper-aware” of risks from Anthropic’s Mythos model and working closely with the company. Reuters-citing coverage later said Goldman had already restricted Anthropic use for Hong Kong employees earlier in the year.

Taken together, the reports point to a tightening circle of controls around frontier AI systems, especially where financial institutions operate across U.S. and Greater China jurisdictions.

What to watch next

The remaining questions are whether JPMorgan has issued a formal internal notice beyond the approved-model list, whether Anthropic updates its public regional-use terms, and whether other major banks follow with similar restrictions outside the United States.

For now, the JPMorgan move adds another sign that access to advanced AI tools is becoming a compliance issue as much as a technology choice.