Xi endorses open-source AI, launches China-led global body

Moonshot AI releases Kimi K3 model hours before speech

SHANGHAI — Chinese leader Xi Jinping endorsed open-source artificial intelligence and called for global AI cooperation at a conference here Friday, according to a Wall Street Journal report. In his speech, Xi warned against “overstretching the concept of national security in the field of AI or placing one country’s security over that of others,” the Journal reported. He did not name the United States.

The Journal reported that Xi positioned China as a champion of openness and that his remarks took aim at U.S. moves to protect its lead in AI semiconductors and models. The speech came as the U.S. and China jockey for leadership of global AI, with the U.S. generally acknowledged on both sides to hold a small lead, according to the Journal. Each country hopes its AI models will become standard around the world, the Journal reported.

MSI previously reported that Chinese officials had held discussions with domestic AI labs about restricting overseas access to the country’s most advanced AI models, a potential shift from Beijing’s previous strategy of promoting open-source models globally.

Xi’s speech was attended by nine Nobel Prize and Turing Award laureates as well as United Nations Secretary-General António Guterres, the Journal reported. Dignitaries from 29 countries had signed up a day earlier to create a new China-led body to promote global AI cooperation.

Xi depicted China as a champion of less-developed nations that fear being left behind as the Trump administration seeks to solidify U.S. dominance in AI, the Journal reported. The new body will “answer the call of the Global South,” he said.

Hours before Xi’s speech, Beijing-based Moonshot AI released an open-weight model called Kimi K3 that has surprised many in Silicon Valley with its advanced capabilities, the Journal reported.

Chinese companies such as DeepSeek, Moonshot and Zhipu AI have published many open-source models that people are generally free to use and adapt, while Silicon Valley firms such as OpenAI and Anthropic have built largely proprietary models using Nvidia’s cutting-edge semiconductors, according to the Journal. The Trump administration recently intervened to temporarily block access to Anthropic’s Mythos model, which can automatically detect cybersecurity flaws, citing national security concerns.

Xi said China would do more to help developing countries with AI training and seminars, and would set up AI-focused bodies within multilateral groupings including the Brics collection of large emerging economies and the Shanghai Cooperation Organization of Central Asian nations, according to the Journal.

“AI development should not be a solo performance by a single country, but a symphony of global collaboration,” Xi said, according to the Journal.

He called for cooperation within the framework of the United Nations to ensure AI is “secure and controllable,” and highlighted the potential dangers of autonomous systems making life-or-death decisions, calling for measures to “ensure that AI is always under human control.”

Kai-Fu Lee, a Taiwan-born AI entrepreneur and venture capitalist who formerly led Microsoft’s and Google’s operations in China, said at the conference that Xi’s nods toward openness would appeal to a world wary of the U.S. further entrenching its global dominance, the Journal reported. Lee said he was skeptical of what he described as Silicon Valley’s “winner-take-all” approach to attaining artificial general intelligence.

“Xi’s message is clear: China is not going to follow anyone on both AI technology and standards,” said George Chen, a partner at The Asia Group who attended the conference, according to the Journal.

Graham Webster, a research scholar at Stanford University who focuses on Chinese tech policy, told the Journal that China’s new AI governance body was more rhetorical than substantive so far. Governing AI risks globally “is by definition going to need the U.S. and China both involved,” Webster said. Otherwise, he said, “it can’t be a comprehensive global solution.”