In a move that highlights the increasingly complex regulatory landscape for global financial institutions operating in Asia, Goldman Sachs has abruptly barred its Hong Kong-based bankers from using Anthropic’s Claude artificial intelligence models. The decision, first reported by the Financial Times on Tuesday, April 29, 2026, and subsequently confirmed by Reuters, marks a significant shift in the bank’s internal technology policy and underscores the growing friction between US-developed AI tools and China’s data governance regime.
According to sources with direct knowledge of the matter, Goldman Sachs employees in Hong Kong previously interacted with Anthropic’s Claude via a proprietary internal AI platform. However, in recent weeks, this access was quietly revoked. The removal of Claude from the Hong Kong platform was not the result of a technical failure but a deliberate policy decision by the bank’s leadership following a consultation with Anthropic regarding the terms of their enterprise contract.
The Compliance Conundrum
The core issue driving Goldman Sachs’ decision appears to be a strict interpretation of its contractual obligations with Anthropic, viewed through the lens of Hong Kong’s unique geopolitical and regulatory status. While AI models like ChatGPT and Claude, built by US firms, are strictly prohibited and geoblocked in mainland China, Hong Kong has historically remained outside these controls, operating under a “One Country, Two Systems” framework. However, usage limits in the territory are increasingly set by US companies rather than local authorities.
An Anthropic spokesperson told the Financial Times that Claude models had never been officially “supported” in Hong Kong, though they declined to comment further on the specific situation with Goldman Sachs. It appears that Goldman, in an abundance of caution, concluded that allowing its Hong Kong-based employees to access Claude, even through an internal, secure platform, could potentially violate the spirit, if not the letter, of its agreement with the AI startup or expose the bank to unforeseen compliance risks related to cross-border data flows.
The Fragmented AI Landscape for Multinationals
The situation at Goldman Sachs illustrates the growing fragmentation of the global AI landscape for multinational corporations. While the bank has removed access to Claude in Hong Kong, other mainstream models, including Google’s Gemini and OpenAI’s ChatGPT, reportedly remain available on Goldman’s internal platform in the territory. The decision to ban Claude did not extend to contracts with other AI vendors, suggesting that the issue is specific to Anthropic’s terms of service or its particular risk assessment regarding the Hong Kong market.
This fragmented approach creates significant operational challenges for global firms. Goldman Sachs Chief Information Officer Marco Argenti stated in February that the bank was actively working with Anthropic to develop AI-powered agents to automate a widening range of internal functions. The inability to deploy these tools uniformly across all global offices, particularly in a major financial hub like Hong Kong, complicates the bank’s broader AI strategy and forces regional teams to rely on a patchwork of different, potentially less capable, technologies.
Hong Kong’s Precarious Position in the Tech War
The Goldman Sachs incident also highlights Hong Kong’s increasingly precarious position in the ongoing US-China technology war. As the United States tightens export controls on advanced semiconductors and scrutinizes the deployment of cutting-edge AI models in jurisdictions perceived as vulnerable to Chinese state influence, Hong Kong finds itself caught in the middle. The territory’s ability to serve as a frictionless bridge between East and West is being severely tested by the rapid advancement of artificial intelligence and the accompanying geopolitical anxieties.
The Hong Kong government and the Hong Kong Monetary Authority did not immediately respond to requests for comment regarding the Goldman Sachs decision. However, local policymakers are undoubtedly aware of the chilling effect such actions could have on the city’s competitiveness as a global financial center. If multinational banks and tech firms increasingly view Hong Kong as a restricted zone for the latest AI tools, it could hinder the territory’s efforts to attract top talent and maintain its status as a premier hub for innovation and finance in Asia.
Looking Ahead: Navigating the AI Divide
As artificial intelligence becomes increasingly central to the operations of global financial institutions, the challenges of navigating the regulatory divide between the US and China will only intensify. The Goldman Sachs decision to bar Anthropic’s Claude in Hong Kong is likely a harbinger of similar actions by other multinational corporations seeking to mitigate compliance risks in a rapidly evolving geopolitical environment.
For AI developers like Anthropic, OpenAI, and Google, the pressure to clearly define and enforce geographic usage restrictions will continue to mount. For global banks, the task of building unified, enterprise-wide AI platforms will require unprecedented agility and a deep understanding of the nuanced and often conflicting, regulatory requirements of the jurisdictions in which they operate. The “One Country, Two Systems” model that has long defined Hong Kong’s economic success is facing a profound stress test in the era of generative AI.
(Related: US Vows Crackdown on Chinese Firms Exploiting AI Models Amid New Legislation)