Kimi: Threat or menace? | TechCrunch

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.

said Moonshot that while Kimi K3 “still follows the most powerful proprietary models, Claude Fable 5 and GPT 5.6 Sol,” the new open source model “demonstrated borderline performance in our evaluation suite, consistently outperforming other models tested.” Independent analyzes of Arena.ai And False AI also suggested that Kimi is competitive with flagship models.

The announcement, which coincided with a speech by Chinese President Xi Jinping at the World AI Conference in Shanghai seems to have scared Wall Street the Nasdaq drops about 1% Friday as investors sold shares in chip companies like Nvidia.

Many of the resulting messages from tech sector figures will sound familiar to those who remember the debate after another Chinese company, DeepSeek, released its open source R1 model in January 2025. Except now, everything seems to have accelerated following the Trump administration’s tariff war with China, repeated battles over the national security threat allegedly posed by Anthropic, and as major AI companies prepare to finally go public.

David Sacks – the Trump administration’s former AI czar and now co-chair of the President’s Council of Advisors on Science and Technology – contrasted Kimi’s progress with a United States “tying itself in knots: politicians and bureaucrats banning new data centers, piling on state regulations, and pushing for new federal agencies to pre-approve border models. That’s how you lose the AI ​​race.” (The news also gave him an excuse do an excavation at Anthropic, calling Claude an example of “woke lobotomized models.”)

And former Uber CEO Travis Kalanick echoed the complaints that the Chinese are ‘distilling’ American AI models (that is, being trained on their results).

“If there is no enforcement against distillation, then everyone should be able to distill from everyone. Otherwise, one arm [would be] tied behind the backs of American models,” Kalanick wrote. (Of course, American models are also built on top of Chinese models, especially Kimi.)

Meanwhile, OpenAI’s head of strategic futures is Dean Ball said that Kimi is “a very good model” whose performance probably cannot be “explained away by distillation or anything like that,” adding that he is “personally surprised that the Chinese state continues to allow the open source of models so well, given the potential risks.”

Ball even suggested that “the likely outcome of a world dominated by an open-weight model is full-blown AI communism,” treating AI as “a ‘public good’ that will ultimately be provided by the state as a kind of ‘digital public infrastructure’.”

“This future seems like a dystopian hell to me, but I’ve never met an open-weight proponent who doesn’t ultimately admit that this is the end of things,” Ball said. He even suggested that the Trump administration (what he worked for) will eventually realize that it must “create large amounts of regulatory risk around the use of Chinese open-weight models.”

“You don’t have to ‘ban’ open source (one of the dumber motifs of the AI ​​policy discussion),” Ball said. “You just have to direct every agency to enact soft laws that create FUD [fear, uncertainty, and doubt]. ‘An Advisory Bulletin from the Federal Reserve shows that there may be backdoors in Chinese AI models.’ It does not have to be so well substantiated. You simply create enough regulatory risk that any regulated company can avoid.”

However, Shakeel Hashim, editor of the AI-focused publication Transformer, argued this Much of the concern is exaggeratedboth because Kimi is “unlikely to have dangerous cyber capabilities,” and because the Chinese government will face “extremely similar incentives” to restrict open Chinese models once they develop those capabilities.