Goldman Sachs Pulls Claude Access for Hong Kong Staff, Sources Say

Goldman Sachs Group Inc. has revoked access to Anthropic's Claude AI for its personnel in Hong Kong, according to people familiar with the matter. The restriction affects an AI agent widely used across financial institutions to accelerate software development workflows. The move signals a tightening of technology controls within one of Wall Street's largest operations in Asia, though neither Goldman nor Anthropic has publicly confirmed the decision or its rationale.

The revocation comes as major banks face mounting pressure to balance productivity gains from generative AI against data security and regulatory compliance obligations. Claude, developed by Anthropic, competes directly with other enterprise AI tools deployed across the financial sector. Goldman has previously explored multiple AI partnerships to modernize trading, risk modeling, and internal operations. The loss of access in Hong Kong specifically raises questions about whether geographic regulatory environments, client confidentiality requirements, or broader institutional risk policies triggered the restriction.

The development highlights the friction between global AI adoption and jurisdiction-specific governance in finance. Firms operating in Hong Kong navigate a complex landscape involving cross-border data flow rules, Chinese regulatory authority, and U.S. compliance obligations. Internal technology bans or limitations in specific markets can signal concerns about intellectual property leakage, competitive intelligence exposure, or regulatory scrutiny that firms prefer to manage quietly. Market observers note that Goldman has not publicly disclosed a formal AI tool restriction policy, making the scope and permanence of this access removal difficult to assess independently.