The US closed access to Claude Mythos, China restricts GLM and Qwen: neural networks have become a strategic weapon

The USA and China synchronized their tightening of control over major AI models in 2026: Washington closed public access to Anthropic's Claude Mythos due to cybersecurity risks, while Beijing is preparing to ban the export of GLM, Qwen, and DouBao — three flagships of China's AI industry.

What Happened in the USA

Anthropic removed the Claude Mythos model from public access, officially citing potential cybersecurity threats. An atypical move: in recent years, the industry has deliberately moved toward greater openness — public APIs, democratized access, open weights for some models.

What exactly in Claude Mythos raised concerns is not officially disclosed. The nature of the restriction suggests that corporate and government clients retain access under separate terms: the model is "removed from public access" rather than fully deactivated. A significant precedent: for the first time, a leading American AI laboratory publicly restricts a top model specifically on national security grounds, not for commercial reasons.

• Claude Mythos is closed to the general public
• Official reason — cybersecurity threats
• Corporate and government clients presumably retain access
• Specific model capabilities that prompted the closure are not disclosed

What China Is Preparing

Beijing is moving on a symmetrical course. Chinese authorities are preparing a mechanism for export control over national AI models: GLM (Zhipu AI), Qwen (Alibaba), and DouBao (ByteDance) will fall under restrictions — the three largest players in China's AI industry.

All three products are actively used outside China. Alibaba's Qwen competes with Western models on key benchmarks and is available through open weights on Hugging Face. DouBao is a mass-market consumer product by ByteDance with hundreds of millions of users. GLM is used in the corporate sector of several countries. Imposing export restrictions will effectively cut off international developers and companies from official access to these tools.

Essentially, this transfers to software the same logic already applied to semiconductors: first, NVIDIA chips came under export controls for China, now the models themselves become subject to bilateral regulation.

"Welcome to the era of digital sovereignty, where neural networks are

officially recognized as strategic weapons," the analysis authors conclude.

What This Changes for the Market

What is happening is not isolated events but part of a systemic shift. Both countries are moving toward a model in which powerful language systems are treated as strategic national security assets — on par with military technology, telecommunications infrastructure, and advanced semiconductors.

For international teams and companies building products based on external AI APIs, this means growing operational uncertainty: a tool available today could be restricted tomorrow without warning. Diversifying your AI stack and being prepared for rapid provider switching cease to be an architectural option and become a production necessity.

In parallel, the configuration of the global AI market is shifting: a single open space is being replaced by several geopolitical zones with different sets of available models, regulatory regimes, and export restrictions.

What This Means

The global AI market is entering a phase of sustained geopolitical fragmentation. The USA and China have created a precedent: the most powerful neural networks are not merely technological products but strategic resources with controlled distribution. Developers and companies worldwide should factor regulatory risks into their architectural decisions now.