To me: Threat or threat? | TechCrunch

Chinese company Moonshot AI released a new version of its Kimi model this week, generating another wave of talk about China and open source AI.

Moonshot said that while the Kimi K3 “still trails the more powerful models, the Claude Fable 5 and GPT 5.6 Sol,” the new open-source model “showed border-line performance throughout our test program, consistently outperforming other tested models.” Independent analysis from Arena.ai and Vals AI also suggested that Kimi competes with flagship frontier models.

The announcement, which coincided with Chinese president Xi Jinping’s speech at the World AI Summit in Shanghai, appeared to rattle Wall Street, with the Nasdaq down nearly 1% on Friday as investors sold shares in chip companies such as Nvidia.

Much of what emerges from tech industry figures will sound familiar to those who remember the debate after another Chinese company, DeepSeek, released its open-source model of R1 in January 2025. Except now, everything seems to be tense after 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.

For example, David Sacks – former AI chief of the Trump administration and now co-chairman of the President’s Council of Advisors on Science and Technology – compared Kimi’s progress to the United States “tying itself in knots: politicians and officials are blocking new data centers, piling on state laws, and pushing new federal agencies to pre-authorize them. (These issues also gave him an excuse to look Anthropic, calling Claude an example of “lobotomized resurrected models.”)

And former Uber CEO Travis Kalanick has echoed complaints that the Chinese are “dumping” (ie, being trained on the effects of) American AI models.

“If the distilling can be forced against, then everyone must be able to distill everyone else … otherwise one arm [would be] tied to the backs of American models,” Kalanick wrote.” (Yes, American models are also based on Chinese ones, especially Kimi ones.)

At the time, OpenAI’s head of future strategy Dean Ball said that Kimi was a “very good model” whose performance “can’t be explained by water smearing or anything like that,” adding that he was “personally surprised that China continues to allow the discovery of models this good, given the potential risks.”

In fact, Ball suggested that “the potential outcome of a country that rules the open-weight model is the full communism of AI,” where AI is treated as a “public good” that will eventually be provided by the government as a form of ‘digital public infrastructure.'”

“This future strikes me as a dystopian hellscape, but I’ve never met an open-weight lawyer who doesn’t ultimately agree that this is where things end,” said Ball. He even suggested that the Trump administration (which used to work for you) will eventually realize that they must “create a greater risk of controlling the use of open-weight Chinese models.”

“You don’t have to ‘ban open source’ (one of the motivations for the AI ​​policy discussion),” Ball said. “You must direct all agencies to issue a soft law that creates FUD [fear, uncertainty, and doubt]. ‘Federal Reserve Advisory Bulletin finds there may be doors to Chinese AI models.’ It doesn’t have to be so well justified. You’re just creating enough regulatory risk that all regulated businesses are going to back off.”

However, Shakeel Hashim, editor of the AI-focused publication Transformer, asserted that many of the concerns are overblown, both because Kimi “probably lacks dangerous cyber capabilities,” and because the Chinese government will face “very similar incentives” to limit China’s open models once it develops those capabilities.