Goldman Sachs has prohibited its Hong Kong bankers from using Anthropic's AI models, reflecting growing concerns about US-China tensions in artificial intelligence. The Wall Street bank implemented this restriction following a strict interpretation of its contract with the Silicon Valley start-up, which concluded that employees in the Chinese territory should not access any Anthropic products, unlike other AI vendors such as OpenAI.

Western AI companies face pressure over potential "distillation" in China, where local actors could train new models through intensive usage of foreign ones. OpenAI previously accused Chinese rival DeepSeek of using its models to develop their own technology, while the White House has accused China of "industrial-scale" theft of US AI intellectual property. No evidence has been disclosed to support these claims.

The restriction poses challenges for Hong Kong's position as a revived financial hub, particularly as bankers use Claude for coding and financial modeling. Hong Kong remains the primary hub for investment banking across Greater China for global banks coordinating cross-border activities. This development could impact other companies with Anthropic enterprise deals operating in the territory.

The crackdown also comes as Anthropic's new Mythos AI model has raised cybersecurity concerns worldwide, with governments and companies worried it could crack current security systems and threaten global financial stability. This move by Goldman signals how AI access restrictions are becoming another front in the broader tech competition between the US and China.