Wall Street's tightening grip on AI access in Hong Kong opens a new front in the technology decoupling between the US and China, as compliance costs rise for global banks operating in the region.
Back
JPMorgan blocks Anthropic AI for Hong Kong staff in 2nd bank move
Wall Street's tightening grip on AI access in Hong Kong opens a new front in the technology decoupling between the US and China, as compliance costs rise for global banks operating in the region.
JPMorgan Chase has restricted its Hong Kong employees from accessing Anthropic's artificial intelligence models, the Financial Times reported Thursday, citing three people familiar with the matter. The move follows a similar action by Goldman Sachs earlier this month, marking the first time major US banks have explicitly blocked staff in the semi-autonomous Chinese territory from using Western AI tools.
"The decision was based on the language in Anthropic's usage terms within its licensing agreement with JPMorgan," the FT report said, without naming the specific clause that triggered the restriction. Staff at the Wall Street bank's Hong Kong office can no longer select Claude models from the internal list of approved large language models, according to the report.
The restriction comes weeks after Anthropic suspended access to its advanced AI model Fable following a US government request to limit access for foreign citizens due to national security concerns. Western AI models including OpenAI's ChatGPT and Anthropic's Claude are already prohibited in mainland China as part of the Great Firewall, though Hong Kong has typically been exempt from such censorship.
The dual actions by JPMorgan and Goldman Sachs suggest a broader compliance recalibration among US financial institutions operating in Hong Kong. Banks face a growing tension between deploying productivity-enhancing AI tools across global offices and adhering to increasingly complex US export controls on advanced technology. JPMorgan's internal approved-LLM list, from which Claude models were removed, is used by staff across its global operations to access AI tools within compliance guardrails.
For Anthropic, the restrictions threaten to undermine its international expansion strategy. The San Francisco-based AI company counts JPMorgan among its enterprise customers. A narrowing addressable market in Asia — where Chinese AI rivals including Baidu's Ernie and Alibaba's Tongyi Qianwen dominate — could pressure the company's revenue growth trajectory outside the US.
The broader implication for the financial sector is clear: AI tool access is becoming a geopolitical compliance issue alongside sanctions and data localization. Banks with significant Hong Kong and China operations, including Citigroup and HSBC, may face similar decisions as US export control rules on AI models continue to evolve. The Treasury Department's proposed rulemaking on outbound investment screening, expected later this year, could further restrict US persons from transferring AI expertise to China-linked entities.
The last time US financial institutions faced a comparable compliance pivot was during the 2020-2021 sanctions escalation against Chinese tech companies, when banks restricted employee access to sanctioned entities' software and cloud services. A similar wave of AI compliance reviews could be on the horizon if more banks follow JPMorgan and Goldman's lead, particularly as the US government tightens scrutiny on AI model transfers to China-linked entities.
This article is for informational purposes only and does not constitute investment advice.
