Anthropic has announced new restrictions that block access to its Claude AI models for entities more than 50% owned by companies in unsupported regions, including China, regardless of where they operate.

The move has created uncertainty for overseas tech tools backed by Chinese firms that rely on Anthropic’s technology.

Trae, a Singapore-based AI code editor launched by ByteDance, uses models from both OpenAI and Anthropic.

Some Trae users are concerned about losing access to Claude and have asked about refunds, though Trae staff said the service is still available for now.

Other Chinese tech companies, including Alibaba’s Qoder and Tencent’s CodeBuddy, also offer products featuring Claude for overseas users.

Chinese AI startup Z.ai has responded by offering incentives to attract users affected by Anthropic’s restrictions.

Anthropic’s decision follows pressure over Chinese firms accessing US AI models via overseas subsidiaries.

In China, domestic AI tools run on local models, as foreign large language models remain unapproved for the market.

.source-ref{font-size:0.85em;color:#666;display:block;margin-top:1em;}

🔗 Source: South China Morning Post

🧠 Food for thought

Implications, context, and why it matters.

Financial strength enables companies to make politically sensitive decisions

Anthropic’s timing reveals how strong financial positioning can enable controversial policy decisions that might otherwise threaten business stability. The company completed a $13 billion funding round just days before announcing the China restrictions, pushing its valuation to $183 billion and demonstrating investor confidence 2. Despite acknowledging the restrictions would impact revenues by hundreds of millions of dollars, Anthropic proceeded with the policy, showing how financial strength creates room for strategic decisions that prioritize long-term positioning over short-term revenue 3.

Ownership-based restrictions represent a new frontier in tech governance

Anthropic’s policy targets entities “more than 50 per cent owned” by companies in unsupported regions, creating a new model that goes beyond traditional geographical restrictions 1. This approach affects overseas operations like ByteDance’s Singapore-based Trae coding tool, demonstrating how ownership structure now matters more than operational location 1. The policy impacts major Chinese tech companies’ international products, including Alibaba’s Qoder and Tencent’s CodeBuddy, forcing them to reconsider their AI technology stack for global markets 1. This represents a shift from previous tech restrictions that focused on where services were used rather than who owned the companies using them, potentially creating a template for other US AI companies facing similar geopolitical pressures.

Recent Anthropic developments

……