Key Takeaways
- President Xi Jinping framed openness as the defining contrast between China’s and the United States’ approaches to artificial intelligence.
- He called for global collaboration in AI development, warning against any single nation monopolising the technology.
- Chinese open‑weight models such as Moonshot’s Kimi K3 and Z.ai’s GLM 5.2 are rapidly closing the performance gap with U.S. frontier models.
- The newly launched World Artificial Intelligence Cooperation Organisation (WAICO) seeks to give the Global South a louder voice in AI governance, positioning itself as a counterweight to Washington’s AI Opportunity Statement.
- While championing open‑source AI, Beijing is simultaneously weighing restrictions on overseas access to its most advanced models, revealing a tension between openness and national‑security concerns.
Opening Remarks on AI Openness
At the World Artificial Intelligence Conference in Shanghai, Chinese President Xi Jinping used the platform’s largest stage to draw a clear line between Beijing’s and Washington’s philosophies on artificial intelligence. He urged nations to seize the “historic opportunity” presented by open‑source AI and cautioned against allowing any single country to dominate the field. Though he never named the United States, the contrast was evident: American labs such as OpenAI and Anthropic keep their frontier models behind closed doors, whereas Xi portrayed China as ready to be more open, to take concrete actions, and to adopt a visionary, cooperative stance. His metaphor—AI development should be a “symphony of global collaboration” rather than a solo performance—captured the core of his message.
Open‑Source as China’s AI Weapon
Xi’s emphasis on openness is not merely rhetorical; it is backed by tangible progress in China’s open‑weight model ecosystem. On the same day he spoke, Beijing‑based Moonshot unveiled Kimi K3, which it claims is the world’s largest open AI model by parameter count. This release followed closely on the heels of Z.ai’s GLM 5.2 and the earlier DeepSeek shock, illustrating how quickly Chinese models are narrowing the gap with U.S. counterparts like ChatGPT, Claude, and Gemini. While American firms have taken a mixed approach—Meta open‑sourcing its Llama series but keeping Muse Spark proprietary, and OpenAI and Anthropic maintaining fully closed models—China’s open‑source push is gaining momentum and credibility as a strategic asset.
Model Advances and Competitive Landscape
The rapid advancement of Chinese open models is reshaping the competitive landscape. Kimi K3’s sheer scale, combined with performance improvements reported by independent benchmarks, suggests that the once‑clear advantage held by U.S. frontier models is eroding. Developers worldwide can now access, modify, and deploy these models without the licensing constraints that characterize many American offerings. This openness lowers barriers for startups, academic institutions, and governments in the Global South, enabling broader experimentation and innovation. Consequently, the strategic value of open‑source AI is increasingly viewed not just as a technical choice but as a geopolitical lever that China is willing to wield.
The New AI Alliance (WAICO) and Its Implications
Behind Xi’s speech lies institutional muscle: a day before his address, Beijing launched the World Artificial Intelligence Cooperation Organisation (WAICO), a 29‑nation coalition headquartered in Shanghai. Members include Indonesia, Brazil, Russia, South Africa, Pakistan, and others from the Global South. Xi described WAICO as a “milestone in the history of world AI development,” envisioning it as a forum where emerging economies can exert greater influence over AI governance, standards, and ethical guidelines. The organisation directly challenges Washington’s AI Opportunity Statement, which has attracted 35 countries. Notably, only Kazakhstan appears on both lists, underscoring that most nations have yet to hedge their bets and are being courted by competing visions of AI leadership.
Balancing Openness with Security
Despite the enthusiastic endorsement of open‑source AI, China’s openness has limits. Reuters reported this month that Beijing is evaluating restrictions on overseas access to some of its top AI models, revealing an underlying tension between the desire to promote global collaboration and the imperative to safeguard national security. Such measures would allow China to retain control over its most sensitive technologies while still benefiting from the reputational and diplomatic advantages of an open‑source posture. This dual approach mirrors strategies employed by other powers, where openness is selectively extended to foster goodwill and adoption, yet core capabilities remain shielded from potential adversaries.
AI Safety and Capacity‑Building Commitments
Xi also devoted considerable attention to AI safety, calling for systems that remain under human control and advocating for early‑warning and emergency‑response mechanisms to prevent loss‑of‑control scenarios. He pledged to provide 5,000 AI training and seminar slots for less‑advanced countries, signalling a commitment to capacity‑building that complements the openness narrative. By coupling safety concerns with offers of technical assistance, China aims to position itself as a responsible steward of AI development—one that seeks to mitigate risks while expanding its influence through education and knowledge sharing.
Geopolitical Stakes of the Open‑Source Divide
The divergence between China’s open‑source advocacy and the United States’ more guarded stance has become a focal point of broader techno‑geopolitical competition. Control over AI standards, access to cutting‑edge models, and the ability to shape global norms are increasingly seen as extensions of national power. Washington’s reliance on closed models allows firms to protect intellectual property and maintain tight usage controls, but it also limits the diffusion of its technology and may alienate partners seeking flexible, cost‑effective solutions. Conversely, China’s open‑source model can accelerate adoption across developing markets, create dependencies on its ecosystems, and amplify its voice in international AI forums—provided it can navigate the security concerns that temper its openness.
Conclusion and Outlook
President Xi Jinping’s speech at the World Artificial Intelligence Conference crystallised a strategic narrative: openness is both a technological advantage and a diplomatic tool for China. By championing open‑source AI, launching WAICO, and offering training opportunities, Beijing seeks to attract allies, especially in the Global South, and to counterbalance U.S. influence in AI governance. Yet the simultaneous consideration of export restrictions reveals that openness will be applied selectively, guided by national‑security imperatives. As the race for AI supremacy intensifies, the interplay between open collaboration and protective safeguards will likely determine which model—China’s symphonic vision or the United States’ more exclusive approach—gains the upper hand in shaping the future of artificial intelligence.