JPMorgan terminates Anthropic access for employees in Hong Kong.

      According to the Financial Times, JPMorgan Chase has prohibited its employees in Hong Kong from utilizing Anthropic’s AI models, a subtle decision that carries significant implications. The bank has taken Anthropic’s Claude off the internal list of approved large language models for staff in the region, leaving them without access to a tool that their colleagues in other locations can still use.

      The report indicates that the reason for this decision was not related to the model’s performance, but rather to the contractual terms in Anthropic’s licensing agreement with JPMorgan. The language used in this agreement led the bank to specifically remove Claude from the approved list in Hong Kong.

      This detail is important: it reflects a contractual and jurisdictional choice, rather than a judgment about the AI model itself, and it restricts this decision to one of the most geopolitically critical financial hubs globally.

      JPMorgan is not the only major bank to take this step; Goldman Sachs had already removed Claude from its list of approved tools available to bankers based in Hong Kong back in April. This action by JPMorgan marks the second instance of a Wall Street firm making a similar move within a short period.

      The context behind this is the increasingly stringent U.S. stance toward advanced AI and its implications for China. Washington has grown more concerned about the potential destinations and users of its most capable AI models, leading to the establishment of stricter policies.

      U.S. Commerce Secretary Howard Lutnick has instructed Anthropic to halt exports of its most advanced models, including those for foreign nationals, due to fears that they might be used for military or intelligence purposes in China, Russia, and other countries deemed a concern.

      For banks, the situation in Hong Kong is complicated. It is both a Chinese territory and a global financial hub, creating compliance challenges when licensing terms introduce uncertainty regarding cross-border data or access. Thus, removing Claude from the approved list is a cautious approach, prioritizing compliance over productivity.

      There is an irony in the timing for Anthropic, as its connections to Wall Street grow even as its access narrows. The company has engaged Morgan Stanley and Goldman Sachs to lead its IPO, with JPMorgan also reported to be involved in the deal, while establishing enterprise partnerships throughout the finance sector. The banks facilitating Anthropic’s entry into the public markets are simultaneously limiting its use in specific offices.

      This contradiction goes beyond IPO underwriting. Over the past year, Anthropic has worked to integrate Claude throughout the financial sector, including a $1.5 billion enterprise joint venture with Wall Street partners and the deployment of the model in banks for accounting and compliance purposes.

      The restriction in Hong Kong does not negate any of these efforts; it simply carves out one jurisdiction from an otherwise broadening relationship. However, it highlights the first instance where geopolitical factors, rather than technology or cost, define where Claude can be employed.

      Additionally, the timing is troubling for Anthropic as it seeks investment from the public markets at a reported valuation in the hundreds of billions, a figure that partly depends on the extent of its enterprise reach. Each bank that restricts Claude in a sensitive market is a small setback to that narrative, illustrating that a frontier AI vendor’s market potential is now influenced as much by export policy as by product quality. The contractual language that led to the restriction has become crucial in a fragmented regulatory environment.

      Neither JPMorgan nor Anthropic has publicly addressed the specifics beyond the Financial Times' report, and it remains uncertain whether the restriction will apply to other regions or if it will be addressed through revised licensing terms. What is clear is this: a second Wall Street bank has removed Claude from its toolkit in Hong Kong, and the motivation points more towards the evolving geopolitical landscape than the technology itself.