July 26, 2026
Can China Keep Its AI Open?
This article was originally published in The Wire China.
On July 16, Chinese lab Moonshot AI released Kimi K3, a 2.8 trillion parameter open-weight model that within a day had displaced Anthropic’s most advanced offering, Fable 5, atop a widely watched frontend coding leaderboard, a first for a Chinese model. The following day, Chinese President Xi Jinping told attendees at the annual World AI Conference in Shanghai to “seize this rare, historic opportunity to encourage open-source.” Yet in the same speech, he warned that AI is “advancing at a staggering speed” and called for “laws and regulations, technological monitoring, early warning and emergency response systems” to keep the technology secure.
Without reliable ways to measure dangerous capabilities, some judgments may be wrong. And an open model that clears a badly drawn line proliferates globally, offering malicious actors anywhere the opportunity to adapt them for harmful uses such as cyberattacks or biological design.
Those two priorities of openness and security will become increasingly difficult to reconcile. As Chinese models approach the frontier, open-weight releases risk diffusing powerful AI capabilities beyond Beijing’s ability to control. Recent reporting suggests Beijing is already beginning to grapple with that dilemma. Nine days before President Xi’s remarks, Reuters revealed that China’s Commerce Ministry had convened Alibaba, ByteDance, and Z.ai to discuss curbing overseas access to their most advanced models, with options ranging as far as barring public release.
Read the full article in The Wire China.
More from CNAS
-
CNAS Insights
Technology & National Security
Washington Can’t Afford to Ignore AI’s Warning ShotOn July 21, OpenAI disclosed the first publicly known incident of AI models escaping their isolated testing environment, gaining unauthorized internet access and hacking into ...
By Ruby Scanlon & Janet Egan
-
Commentary
Technology & National Security
The Modern Mythology of AILLMs can’t replace people, and AI can’t replicate actual thought. But if employers reject these truths and succumb to empty promises and motivated thinking, employees will suf...
By Peter L. Levin
-
Video
Technology & National Security
Did the U.S. Just Lose Its AI Advantage to China?For years, the United States has tried to slow China's technological rise by restricting access to advanced chips and semiconductor technology. But new trade figures suggest C...
By Daniel Remler
-
Commentary
Technology & National Security
The Kill Switch and the Long ArmBut much remains undecided: The EU’s rules are still being written, and the biggest investments are still only plans. An assurance offered now can shape those choices but if o...
By Pablo Chavez
