OKX Restricts Claude Use in Hong Kong Over AI Access Rules

OKX has restricted some employees in Hong Kong from using Anthropic’s Claude artificial intelligence tools after the cryptocurrency exchange’s enterprise account was temporarily suspended earlier this month.

The restriction also applies to employees traveling through mainland China, according to information reported by Bloomberg and highlighted by blockchain news account @WuBlockchain on X. The enterprise account has since been restored, but OKX said some previous usage may not have complied with Anthropic’s regional access requirements.

The development highlights the growing challenges facing global technology and cryptocurrency companies as they increasingly rely on artificial intelligence while operating across jurisdictions with different rules for AI services.

OKX has become an active user of AI technology internally, with the company encouraging employees to incorporate artificial intelligence into their daily work. The exchange reportedly spends between $6 million and $8 million every month across major AI model providers.

OKX Redirects AI Requests to Other Models

Following the issue with Claude access, OKX said it would redirect relevant AI requests to other available models for employees in affected locations.

The decision allows the company to continue using artificial intelligence while reducing its dependence on a single provider in regions where access restrictions may apply.

Anthropic's current regional access policy does not permit Claude to be used in mainland China or Hong Kong. The restrictions create an additional challenge for international companies that operate offices or employ staff across multiple markets.

For OKX, the situation appears to have been handled primarily through internal restrictions rather than a complete reduction in its use of AI.

The exchange has continued to emphasize the importance of artificial intelligence in its operations, making access to alternative models an important part of its strategy.

OKX Reportedly Spends Up to $8 Million on AI

The scale of OKX's AI spending underscores how quickly artificial intelligence has become an important business expense for large technology-driven companies.

OKX said it spends approximately $6 million to $8 million per month across major AI model providers.

That translates to potentially tens of millions of dollars in annual spending on AI infrastructure and services.

The spending reflects the increasing use of large language models and other AI technologies for tasks ranging from software development and research to internal productivity and business operations.

Rather than treating AI as an experimental technology, OKX appears to be incorporating it more deeply into its corporate structure.

The company has also reportedly integrated AI capabilities into employee performance evaluations, encouraging workers to adopt AI tools as part of their responsibilities.

AI Becomes Part of Employee Performance

OKX's approach illustrates how companies are changing workplace expectations as AI becomes more widely available.

Employees are increasingly being evaluated not only on traditional productivity measures but also on their ability to use AI tools effectively.

For a global cryptocurrency exchange, the potential applications are broad.

AI can assist with software development, data analysis, customer support, research, compliance work and internal documentation. It can also help employees process large amounts of information more efficiently.

However, widespread AI adoption also creates new challenges, particularly around data security, privacy and access restrictions.

Companies must ensure that employees use AI services in accordance with the policies established by the providers and the regulations applicable in each markets.

The Claude incident demonstrates how complicated that task can become when employees work across different jurisdictions.

Claude Account Was Temporarily Suspended

The issue reportedly began after OKX's enterprise account with Anthropic was briefly suspended in early August.

The account has since been restored, but the incident prompted the exchange to review how employees were accessing Claude.

OKX acknowledged that some previous usage may not have complied with Anthropic's regional access policies.

The company subsequently moved to restrict Claude access for employees in Hong Kong and for staff traveling through mainland China.

The move appears designed to prevent further policy conflicts while allowing OKX employees to continue using other AI systems.

The situation also demonstrates that enterprise access to AI models is subject to the policies of individual providers. Even large companies can face restrictions if usage does not align with a provider's geographic or contractual requirements.

Regional AI Restrictions Create Challenges

AI companies have increasingly established geographic restrictions around access to their products.

These policies can be influenced by regulatory requirements, business decisions, security concerns and the availability of AI services in particular jurisdictions.

For multinational companies such as OKX, regional restrictions can create operational complications.

An employee may normally work in a jurisdiction where an AI service is available but temporarily travel to a location where the same service is restricted.

Without appropriate controls, that employee could unintentionally violate the provider's access requirements.

OKX's decision to redirect AI requests to other models provides one potential solution. It allows employees to continue using AI while the company manages geographic restrictions at the organizational level.

Cryptocurrency Industry Increasingly Embraces AI

The development comes as artificial intelligence and cryptocurrency increasingly overlap.

Crypto companies are using AI for a growing number of functions, including trading research, blockchain analysis, security monitoring, customer service and software development.

The combination of AI and blockchain has also become a major investment theme, with companies exploring autonomous agents, decentralized computing and AI-powered financial applications.

For major exchanges, AI can potentially improve internal efficiency while helping companies process the enormous amount of data generated by digital asset markets.

However, increased adoption also means companies must pay greater attention to model reliability, data privacy and regulatory compliance.

OKX's reported monthly AI spending demonstrates the scale at which large cryptocurrency companies are now investing in these technologies.

What the OKX Incident Means for Corporate AI Use

The situation involving Claude and OKX highlights an emerging issue for companies that depend heavily on third-party AI providers.

AI tools are becoming deeply integrated into everyday business operations, but access to those tools is not always universal.

Companies operating internationally must consider not only the capabilities and costs of AI models but also where those models can legally or contractually be accessed.

For OKX, maintaining access to multiple AI providers could help reduce the impact of future restrictions or service interruptions.

The company's decision to redirect affected requests suggests that AI diversification may become increasingly important for large organizations.

As artificial intelligence becomes a larger part of corporate operations, companies will likely need more sophisticated systems for managing which employees can access which models and from which locations.

For OKX, the temporary Claude suspension appears to have accelerated that process, while its reported $6 million to $8 million monthly AI budget shows just how strategically important artificial intelligence has become to the cryptocurrency exchange.