Open Models Are Becoming a Tool of Chinese Statecraft

The argument in Washington over the past few months has been about whether to ban Chinese AI models or restrict them. What has been less examined is what Beijing intends to do with the open-weight – that is, publicly available and downloadable – models its labs have built. Beijing has started to answer that question, and not with one voice.

On 16 July, the UN secretary-general attended as 29 countries in Shanghai signed the founding agreement of the World Artificial Intelligence Cooperation Organization (WAICO), which will be headquartered in the city.

What China brought to the Shanghai meeting seemed less like a plan than a possession: the largest collection of open-weight AI models in the world. These are models that are already lodged in the infrastructure of countries that compete with China. What July gave this collection was a political sponsor and an institution to carry it abroad. An institution that can spread a resource, though, can also meter it.

The second tendency runs in a direction not many people expected. The weights enter advanced democracies. Countries without indigenous frontier-model labs must build on somebody else’s foundation, and increasingly these foundations are Chinese open-weight models. Rakuten AI 3.0, launched in March as Japan’s largest model and funded in part by Geniac, the government’s national AI programme, turned out, on inspection of its configuration files, to be built on DeepSeek V3. Another Geniac model, created by the lab Abeja and also funded by the Japanese government, is built on Alibaba’s Qwen. Singapore’s flagship model, Sea-Lion, is also built on Qwen.

Seen from Beijing, the logic is not hard to follow. Under US export controls and company-blacklisting measures designed to slow China down, the open-weight route has worked. It generates domestic innovation, it allows Chinese models to travel abroad without an export licence and it gives China a lever on AI infrastructure that others are building.

On paper, WAICO has no open-source mandate. But look at what it is set up to deliver.

Chinese President Xi Jinping’s keynote speech at the founding of WAICO cast China as a provider of international public goods in AI and, in one line, encouraged ‘open source’ and ‘openness’. The National Development and Reform Commission issued an action plan the same day, with eight priority areas. The third is open-source ecosystem sharing, which includes the joint development of open-source compliance systems and security norms. It also promises to support localised innovation abroad built on open models.

The conference chair’s statement calls for ‘open and inclusive international open-source communities’. Two days later, Chinese state media Xinhua reported what the organisation would actually pursue. Capacity building for developing countries, such as training, policy exchange and joint research, would come first, followed by a platform matching countries that have AI technology with countries that want it and, once again, encouragement of jointly built open-source ecosystems.

Xi’s speech also put money behind the project, with 5,000 training places for developing countries over five years and cooperation centres aimed at the Association of Southeast Asian Nations, the Arab League, the African Union, Latin America, the Shanghai Cooperation Organisation and BRICS. Read the founding documents and you learn little. Read the delivery list and you find that China’s open models have acquired a policy platform for their diffusion abroad, with a founding agreement, a headquarters and 29 governments behind it.

All of this looks like a gift to the world. But gifts from governments often come with terms. Treating openness as strategy, China must decide how much to permit, to whom, and on what terms. Deciding how to be open is, in practice, the same exercise as deciding how to restrict.

The signals so far suggest Beijing has started that exercise, and they point in different directions. While Xi’s one line on open source carried no conditions, the conference chair’s statement attaches three: open-source ecosystems should be built in a responsible manner, with respect for corporate autonomy and on the premise that intellectual property is protected. These are only mild conditions.

An AI safety governance framework issued in September 2025 by China’s national cybersecurity standards committee urges open-source model providers and communities to refine their release rules and to spell out prohibited uses of downloaded models. A national standard on open-source model security is now in preparation, covering stages including development, release, re-release, community operation and deployment.

China’s open models grew out of commercial choices. They now have a government minding them and an organisation to carry them abroad. That sponsorship could make them more stable or more conditional, and it is too early to say which.

Kai-Shen Huang is an assistant research professor at the Institutum Iurisprudentiae (Institute of Law) of Academia Sinica, Taipei.

Image of Chinese Foreign Minister Wang Yi and UN Secretary General Antonio Guterres among other officials at the founding meeting of WAICO in Shanghai on 16 July 2026: Chinese Ministry of Foreign Affairs.

AI contributed no ideas to this article – Kai-Shen Huang.

This article appeared originally at The Strategist (ASPI).