Wall Street Reacts to Burnham’s Fiscal Plans

Anthropic has implemented stricter measures to block users in mainland China from accessing its Claude artificial intelligence models, according to a July 3, 2026, report from the Financial Times. The company is tightening controls to ensure its AI technology remains unavailable in the region, targeting technical workarounds used to bypass existing geographic restrictions.

Why is Anthropic restricting Claude in China?

The restrictions align with broader United States national security objectives to prevent the transfer of advanced AI capabilities to Chinese entities. According to the Financial Times, the move is part of a wider effort to maintain a technological lead in frontier AI models.

This software-level restriction mirrors hardware policies established by the U.S. Department of Commerce. The department has previously restricted the export of high-end semiconductors, such as those produced by Nvidia, to prevent China from training large-scale AI models that could have military applications.

By clamping down on access to the finished model, Anthropic prevents Chinese developers and state-linked firms from using Claude’s reasoning and coding capabilities to accelerate their own domestic AI development or automate cyber operations.

How is the access being restricted?

While Anthropic has not officially supported China for some time, users frequently accessed Claude through Virtual Private Networks (VPNs) and third-party API aggregators. The Financial Times reports that the company is now actively identifying and blocking these access points.

Technical enforcement typically involves several layers of verification:

• Monitoring API traffic for patterns consistent with Chinese data centers or known VPN exit nodes.
• Implementing more rigorous identity verification for new account creations.
• Revoking API keys associated with “wrapper” services that resell Claude access to users in restricted jurisdictions.

These measures target the “gray market” of AI access, where third-party developers create interfaces that act as bridges between restricted users and the model’s backend.

How does this compare to other AI companies?

Anthropic’s approach follows a precedent set by OpenAI and Google. OpenAI restricted ChatGPT access in mainland China shortly after its launch, though it struggled for years to eliminate the proliferation of “mirror sites” that provided cached or proxied versions of the tool.

Google has similarly kept its Gemini models unavailable in China. However, the Financial Times suggests that Anthropic’s current “clamp down” indicates a more aggressive technical posture toward enforcement than earlier attempts by its peers.

The contrast in strategy is evident when comparing the “soft blocks” of 2023 and 2024—which relied mostly on IP blocking—to the 2026 efforts that target the underlying API ecosystem and account verification processes.

What are the business implications for AI developers?

The crackdown creates a significant hurdle for Chinese startups that have integrated Claude into their software workflows. Many of these firms relied on API aggregators based in the U.S. or Europe to maintain their services.

The sudden loss of access forces these companies to pivot to domestic alternatives. This shift benefits Chinese AI firms like Baidu and Alibaba, which offer competing large language models (LLMs) designed to comply with local regulations and avoid U.S. export risks.

For Anthropic, the move carries a trade-off. While it secures the company’s standing with U.S. regulators, it forfeits potential revenue from one of the world’s largest technology markets. The decision suggests that regulatory compliance and national security concerns now outweigh the financial incentive of expanding the user base in China.

Industry analysts note that this trend of “digital decoupling” is likely to accelerate as AI models become more capable of autonomous coding and strategic planning, increasing the perceived risk of their misuse by foreign adversaries.