Key Points:
- Goldman Sachs recently removed access to Anthropic’s Claude artificial intelligence for its bankers stationed in Hong Kong.
- The Wall Street bank made this strict decision after a close review of its service contract with the technology startup.
- Despite the ban on Claude, Hong Kong employees can still access other popular AI models, such as Google Gemini and OpenAI’s ChatGPT.
- The restriction follows a major February announcement where Goldman Sachs revealed plans to automate tasks using Anthropic technology.
Goldman Sachs has officially restricted its employees in Hong Kong from accessing Anthropic’s popular Claude artificial intelligence model. The sudden move surprised many bankers who had previously used the system daily to speed up their work. The decision highlights the growing tension and complicated legal rules surrounding how American technology companies operate their newest tools in specific global regions.
According to a source with direct knowledge of the situation, the change happened quietly over the last few weeks. Before the restriction, bankers located in the Chinese territory could easily open the bank’s 1 main internal artificial intelligence platform and ask Claude for help. Employees used the digital tool to organize complex data, draft long emails, and summarize financial documents. Now, the internal system completely denies them access to the Claude interface.
The Financial Times first reported the removal of access on Tuesday, citing several people familiar with the matter. The report explained that Goldman Sachs made this choice after holding a serious consultation with Anthropic. Following their talks, bank executives reviewed their service contract and decided to take an extremely strict interpretation of the rules. They firmly concluded that staff working inside Hong Kong should not use any Anthropic products at all.
An Anthropic spokesperson later confirmed some details regarding their geographical rules. The representative stated that the company never officially supported the Claude models for users in Hong Kong. However, the spokesperson declined to share any further details or comment on their private talks with the investment bank.
This situation highlights the confusing rules governing access to artificial intelligence in Asia. The United States government currently prohibits American companies from offering advanced AI models in mainland China. However, Hong Kong usually falls into a strange gray area regarding technology rules. Usage limits in the city often depend entirely on the American technology companies themselves, rather than strict government bans or international trade laws.
Despite losing access to Claude, Goldman Sachs staff in Hong Kong still have several digital options to help them with their daily tasks. The source confirmed that employees can still reach other mainstream artificial intelligence tools. The bank’s central platform continues to offer OpenAI’s ChatGPT and Google’s Gemini to the Hong Kong teams. The new restriction only applies to the specific contract signed with Anthropic.
So far, leaders involved in the situation have chosen to stay quiet. Goldman Sachs declined to offer any official comment to reporters regarding their internal software rules. Likewise, the Hong Kong government and the Hong Kong Monetary Authority did not immediately respond to requests for an explanation about how artificial intelligence tools operate within their borders.
This unexpected block comes at an interesting time for the famous Wall Street bank. Just a few months ago, in February, Goldman Sachs Chief Information Officer Marco Argenti proudly announced a major partnership with Anthropic. He stated that the bank wanted to build fully autonomous digital agents powered by the Claude model. For 6 straight months, Anthropic software engineers worked closely with the bank to develop these new tools.
Argenti planned for Claude to act exactly like a digital coworker for the bank’s massive accounting and compliance departments. The goal was to automate thousands of boring internal tasks. In fact, bank leaders noted that artificial intelligence can now draft up to 95% of standard financial filing documents in just a few minutes. The bank opened its internal artificial intelligence assistant to all 46,500 global employees, hoping to make the entire company much faster and more efficient.
It remains completely unclear how the new Hong Kong restriction will alter the bank’s overall technology goals. Because Goldman Sachs operates globally, maintaining different software tools across regions could create extra work for its engineering teams. However, with systems like Gemini and ChatGPT still working perfectly fine, the Hong Kong office will simply shift its daily automation tasks to those other available programs instead.
