China's open-weight AI a boon for the world

The 2026 World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance, hosted by Shanghai earlier this month, was themed "AI Partnership for a Brighter Future" and focused on the global development and governance of AI technology.

Currently, China and the United States are the only two nations with the resources to develop AI technology on a large scale. Therefore, their approaches to AI development significantly shape the global trajectory of AI evolution. While one country emphasizes AI as a tool accessible to all, the other has adopted a model where AI is controlled by the elite.

The success of the two approaches will depend not only on the capabilities of individual models but also on their global adoption. In recent months, the performance gap between Chinese and US frontier models has narrowed significantly. Moonshot's Kimi K3, for instance, recently achieved benchmark results rivaling those of Anthropic's Claude Opus 4.8 and OpenAI's GPT-5.5.

However, the more consequential divergence lies in governance philosophy. While US companies often retain proprietary control, China's leading labs and tech giants have embraced open-weight models. This means their model parameters are publicly accessible for modification and deployment, enabling global adaptation and innovation. Although this is not without the risks of misuse, it offers a viable alternative to AI monopolies.

The WAIC's emphasis on openness signals that this ethos is now central to Beijing's AI vision. As President Xi Jinping stated on July 17 in his address at the WAIC, "AI development should not be a solo performance by a single country, but a symphony of international cooperation. We should seize this rare, historic opportunity to encourage open source, openness, collaboration and sharing."

This message is like a beacon to the Global South, indicating that this powerful technology will not remain under the control of US tech giants. It allows developers from Southeast Asia, Africa, and Latin America to finetune Chinese models for their own languages and contexts while avoiding high costs. Countries that were previously limited to consuming technology can now actively participate in shaping its development.

Thanks to this ethos, since August 2025, Chinese AI has surpassed Western AI in terms of cumulative downloads in the open-source community, highlighting its widespread adoption across the globe. Many startups and companies, including those from the US, are increasingly using Chinese large language models to build their own AI programs.

For the world to enter the AI age equitably, there must be equal access to this technology. Otherwise, it could become another tool of hegemony and domination.

We have already observed that in Australia, where US companies are embedding themselves into the process of legislating AI and data center standards. The Australian government is resisting these influences, hoping to put in place measures to support Australian artists and the energy grid, but the threat to Australian digital sovereignty persists.

Recent moves by the US government to possibly ban US companies from using Chinese open-source models highlight its apprehensions concerning Chinese AI — not just its performance, but how ubiquitous it can potentially become.

Similarly, measures are being considered to ensure that US AI stays in the hands of US users. In June, the US government went so far as to briefly ban Claude's highest-performing model, Fable 5, from use by anyone who is not a citizen of the US.

However, this pursuit of absolute control over AI may lead to its isolation from global AI development, ultimately weakening its position in the long run.

China's embrace of open-weight models in AI is a boon to the world. It will ensure that AI is accessible to all, and does not remain in the hands of corporate tech giants.

Without this ethos, AI is likely to become another divide between developing and developed nations — another form of fragmentation that breeds inefficiency and mistrust.

If AI is to become a force for good, then it must be developed through multilateral cooperation, allowing all stakeholders to guide the technology and the changes it will bring.

The author is a research affiliate at the AustChina Institute, a Melbourne-based foreign policy think tank.

The views don't necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.