China sought access to Anthropic's newest artificial intelligence models, but the San Francisco-based AI company declined, according to a report published Monday by The New York Times — a development that highlights how the most advanced AI systems have become a new flashpoint in the intensifying technological competition between Washington and Beijing.
China made efforts to obtain access to Anthropic's latest AI models, only to be turned away, according to reporting by Dustin Volz, Julian E. Barnes, Sheera Frenkel and Tripp Mickle in The New York Times. The refusal reflects growing concerns in Washington and among leading AI companies about the national security implications of transferring cutting-edge AI capabilities to foreign adversaries.
Anthropicis among the most prominent players in the current wave of advanced AI development, alongside OpenAI. According to the Times report, the newest models from both companies are helping extend the United States' technological lead over China in artificial intelligence — a sector that both governments have identified as strategically vital.
A Contested Technology Landscape
The episode is the latest sign that frontier AI models — systems capable of complex reasoning, coding, scientific analysis and more — are increasingly treated as sensitive national assets rather than commercial products freely available on the open market.
The US government has moved in recent years to restrict China's access to advanced semiconductors and related technologies through export controls, arguing that such tools could be used to accelerate military capabilities or enable authoritarian surveillance. AI model access is emerging as the next front in that effort.
Anthropichas not made an extensive public comment on the specific incident as reported. The company has, however, previously been vocal about the need for AI safety and has engaged with US policymakers on questions of national security.
A Widening Rivalry
China has invested heavily in its own domestic AI industry, with companies such as Baidu, Alibaba and the startup DeepSeek making notable advances. Earlier this year, DeepSeek's release of a capable, low-cost AI model drew significant attention, raising questions about how wide the gap between American and Chinese AI capabilities truly is.
Nonetheless, US officials and AI researchers have maintained that the very top tier of American AI systems — such as Anthropic's Claude models and OpenAI's GPT-4 series — remain ahead of publicly available Chinese equivalents, particularly in certain reasoning and multimodal tasks.
The refusal to grant China access to Anthropic's models signals that leading AI companies are now operating as de facto participants in US technology policy, making consequential decisions about who can access their most powerful systems.
Analysis
Why This Matters
- Access to frontier AI models is increasingly treated as a national security issue, meaning private companies are now key gatekeepers in US-China tech rivalry — a role with enormous geopolitical implications.
- If the US successfully restricts China's access to top-tier AI systems, it could slow China's ability to apply those tools to military, intelligence and economic applications — but may also accelerate China's push to build fully independent domestic alternatives.
- The incident sets a precedent for how AI companies handle access requests from foreign state actors, potentially influencing future regulatory frameworks around AI export controls.
Background
The United States and China have been locked in a broadening technology competition for several years. Beginning under the Trump administration and continuing under Biden and into the current period, Washington has imposed sweeping export controls on advanced semiconductors — particularly Nvidia chips — to prevent China from acquiring the hardware needed to train and run powerful AI models at scale.
Anthropicwas founded in 2021 by former OpenAI researchers, including Dario and Daniela Amodei, and has grown into one of the most closely watched AI companies in the world. Its Claude series of models competes directly with OpenAI's GPT systems and Google's Gemini. The company has received significant investment from Amazon and has been active in engaging with US policymakers on AI governance.
China, meanwhile, has pursued an aggressive domestic AI strategy, pouring state funding into research and incentivising private companies to close the gap with American rivals. The release of DeepSeek's R1 model in early 2025 briefly rattled markets, suggesting China was advancing faster than many had assumed — though US officials argued the top American models retained meaningful advantages.
Key Perspectives
US AI Companies (Anthropic): By declining China's access request, Anthropic is effectively aligning itself with US national security interests, signalling that frontier AI capabilities will not be treated as ordinary commercial goods available to any paying customer.
Chinese Government and Industry: Beijing has consistently argued that US technology restrictions amount to unfair economic competition designed to contain China's rise. Chinese officials and companies are likely to accelerate self-sufficiency efforts in response to such refusals.
Critics and Sceptics: Some researchers caution that restricting model access may be less effective than assumed — determined state actors can attempt to replicate capabilities through independent research, and overly broad restrictions risk fragmenting the global AI ecosystem without meaningfully improving security.
What to Watch
- Whether the US government moves to formalise AI model export controls, extending existing chip restrictions to cover access to frontier model weights or APIs.
- Anthropic's and OpenAI's next public statements on their policies regarding foreign government access requests.
- Progress from Chinese AI labs on closing the capability gap, which will determine how long US restrictions remain strategically meaningful.