Anthropic restricts Claude access in Hong Kong, affecting users and firms

Anthropic, a US artificial intelligence company, has reportedly intensified enforcement of geographical restrictions on its Claude AI models, leading to account suspensions for users in Hong Kong who accessed the service via virtual private networks (VPNs). Reports of these suspensions surfaced on social media platforms X and Reddit on Thursday, with users who had been using VPNs for months finding their accounts suddenly disabled.

Similarly, crypto exchange OKX has restricted Claude access for its employees in Hong Kong and those traveling through China. This action followed a brief suspension of OKX's own corporate account for Claude earlier this month. OKX stated that some employees may have used Claude in ways that did not fully comply with Anthropic’s geographic access policies, which do not permit Claude's use in Hong Kong or mainland China. The exchange indicated it would redirect affected requests to other AI models.

OKX's monthly spending on AI providers ranges from $6 million to $8 million, and the company encourages AI integration into daily work. This multi-model approach allows OKX flexibility as AI providers implement varying regional access rules.

This situation is not unique to OKX. Goldman Sachs and JPMorgan have also previously restricted Claude access for their Hong Kong-based employees due to licensing agreements that did not cover the territory. The issue highlights the complexities multinational companies face in accessing US-developed AI systems, with employee software access varying based on physical location. For Hong Kong, a major international financial center and crypto hub, these restrictions are particularly significant.