Token diplomacy: How China is shaping the AI future

5 hours ago

At the UN AI for Good summit in Geneva, Chinese officials promoted free open-source AI models to developing nations, contrasting with the absence of US frontier lab leaders and Trump administration officials.
China's AI strategy mirrors its previous playbook with solar panels and EVs: flooding the world with cheaper models to set de facto standards and gain market dominance.
Beijing's 'token diplomacy' shifts focus from infrastructure (Belt and Road) to AI tokens, offering open-source models at lower costs and training for global adoption.
Some countries remain skeptical due to Chinese AI's opacity and potential dependency, while others, like Tanzania, are developing sovereign models using both Chinese and US AI.
Xi Jinping called for open-source AI and opposed overstretching national security concepts, while the US Treasury secretary warned against IP theft through model distillation.
Room for disagreement: Entities like Singapore and Aramco embrace Chinese AI, and critics argue Western models also have embedded biases, while Chinese open weights allow user customization.

OpenRouter aggregates AI models from various providers, enabling developers to choose efficient options.
Chinese models, particularly DeepSeek, are rising in popularity on OpenRouter, reflecting a shift in US companies' AI access.
Flo Crivello notes a potential Chinese cutoff would impact firms but not be catastrophic, with US companies needing 3-6 months to match Chinese open-weight models.

Model distillation is increasingly seen as a national security concern, with Chinese AI firms like Zhipu accused of extracting capabilities from U.S. frontier models.
Mira Murati's Thinking Machines built its first foundation model using Chinese models such as DeepSeek-V3 and Moonshot AI's Kimi K2.5, highlighting reciprocal innovation.
Distillation is a common machine learning technique, not new or inherently nefarious, but now politically charged in the context of U.S.-China AI competition.

DeepSeek's open-source R1 model released in January 2025, which rivals top US AIs and is free, caused a $1 trillion drop in US tech stocks and prompted US lawmakers to propose banning it on government devices.
Z.ai's GLM-5.2 release in contrast did not cause similar panic, indicating a shift in the US-China AI arms race where China now catches up, similar to other technologies like smartphones and electric cars.
Chinese AI models are typically open-source and free for local use, while US models are often cloud-based, fee-charging, and closed, leading to differences in accessibility and innovation strategies.

US AI strategy relies on proprietary, locked-down models, while China adopts an open-weights approach.
AI models lack strong moats; switching between them is easy via APIs, making enterprise services the real differentiator.
China's open models exploit US export controls by turning compute disadvantages into distribution advantages and commoditizing the AI layer.

Chinese President Xi Jinping urged international cooperation on AI at the World Artificial Intelligence Conference, opposing single-country dominance and stressing human control.
China launched the World Artificial Intelligence Cooperation Organisation (WAICO) with 29 founding nations, aiming to shape global AI regulations and promote safe, beneficial technology.
Analysts suggest Beijing uses WAICO to influence UN AI policies, positioning China as a leader in the Global South amid tech competition with the U.S.

OpenRouter aggregates AI models from various providers, enabling developers to choose efficient options.
Chinese models, particularly DeepSeek, are rising in popularity on OpenRouter, reflecting a shift in US companies' AI access.
Flo Crivello notes a potential Chinese cutoff would impact firms but not be catastrophic, with US companies needing 3-6 months to match Chinese open-weight models.

Model distillation is increasingly seen as a national security concern, with Chinese AI firms like Zhipu accused of extracting capabilities from U.S. frontier models.
Mira Murati's Thinking Machines built its first foundation model using Chinese models such as DeepSeek-V3 and Moonshot AI's Kimi K2.5, highlighting reciprocal innovation.
Distillation is a common machine learning technique, not new or inherently nefarious, but now politically charged in the context of U.S.-China AI competition.

DeepSeek's open-source R1 model released in January 2025, which rivals top US AIs and is free, caused a $1 trillion drop in US tech stocks and prompted US lawmakers to propose banning it on government devices.
Z.ai's GLM-5.2 release in contrast did not cause similar panic, indicating a shift in the US-China AI arms race where China now catches up, similar to other technologies like smartphones and electric cars.
Chinese AI models are typically open-source and free for local use, while US models are often cloud-based, fee-charging, and closed, leading to differences in accessibility and innovation strategies.

US AI strategy relies on proprietary, locked-down models, while China adopts an open-weights approach.
AI models lack strong moats; switching between them is easy via APIs, making enterprise services the real differentiator.
China's open models exploit US export controls by turning compute disadvantages into distribution advantages and commoditizing the AI layer.

Chinese President Xi Jinping urged international cooperation on AI at the World Artificial Intelligence Conference, opposing single-country dominance and stressing human control.
China launched the World Artificial Intelligence Cooperation Organisation (WAICO) with 29 founding nations, aiming to shape global AI regulations and promote safe, beneficial technology.
Analysts suggest Beijing uses WAICO to influence UN AI policies, positioning China as a leader in the Global South amid tech competition with the U.S.

Hasty Briefsbeta