Chinese company Moonshot AI released a new version of its Kimi model this week, sparking a new wave of discussion about China and open source AI.
Moonshot said that while Kimi K3 “still falls short of the most powerful proprietary models, Claude Fable 5 and GPT 5.6 Sol,” the new open source model “demonstrated frontier-level performance across our evaluation suite and consistently outperformed other models tested.” Independent analysis by Arena.ai and Vals AI also suggests Kimi can compete with flagship Frontier models.
The announcement, which coincided with Chinese President Xi Jinping’s speech at the World AI Conference in Shanghai, appeared to surprise Wall Street, with the Nasdaq falling about 1% on Friday as investors sold shares in semiconductor companies such as Nvidia.
Many of the posts from tech industry insiders will sound familiar to anyone who remembers the controversy after another Chinese company, DeepSeek, released its open source R1 model in January 2025. Everything seems to be on the rise, except for the Trump administration’s tariff war with China, repeated battles over the alleged national security threat posed by Anthropic, and now that the big AI companies are finally preparing to go public.
For example, David Sachs, former Trump administration AI czar and current co-chair of the President’s Council of Advisors on Science and Technology, contrasted Kimi’s progress with the United States, which is “in a bind. Politicians and bureaucrats are banning new data centers, piling on state regulations, and requiring new federal agencies to pre-approve frontier models. That’s how we lose the AI race.” (This news also gave him an excuse to criticize Anthropic, calling Claude an example of a “woke lobotomy model.”)
Former Uber CEO Travis Kalanick also echoed complaints that the Chinese are “distilling” (i.e., being trained on) American AI models.
“If distillation were not forced, then everyone should be able to distill from everyone else. Otherwise, the American model would have one arm tied behind its back,” Kalanick wrote. (Of course, the American model is also based on the Chinese model, especially Kimi.)
Meanwhile, Dean Ball, head of strategic futures at OpenAI, said Kimi is a “very good model” whose performance probably “can’t be explained by distillation or anything,” adding: “I’m personally surprised that the Chinese state continues to allow such a good model to be open sourced given the potential risks.”
Indeed, Ball suggested that “a possible outcome of a world where the open-weight model prevails is full-blown AI communism,” where AI is treated as “a ‘public good’ that will ultimately be provided by states as a kind of ‘digital public infrastructure.'”
“This future strikes me as a dystopian hellscape, but I’ve never met a proponent of the open weight model who doesn’t finally accept that this is the end of things,” Ball said. He even suggested that the Trump administration (for which he previously worked) would eventually recognize the need to “create significant regulatory risk around the use of China’s open-grade model.”
“There’s no need to ‘ban open source’ (one of the silly motifs of AI policy discussions),” Ball said. “Just tell your agencies to issue soft laws that create FUD (Fear, Uncertainty, Doubt).” “Federal Reserve Advisory Bulletin Finds China’s AI Models May Have Backdoors.” It doesn’t need much justification. It would only create enough regulatory risk that all regulated companies would back away. ”
But Shakir Hashim, editor of the AI-focused publication Transformers, argued that many of those concerns are overblown, as Kimi “is unlikely to have dangerous cyber capabilities” and the Chinese government faces “very similar incentives” to limit the open China model if it develops those capabilities.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
