\n<\/p>\n
Chinese company Moonshot AI released a new version of its Kimi model this week, generating another wave of discourse about China and open source AI.<\/p>\n
Moonshot said<\/a> that although Kimi K3 \u201cstill trails the most powerful proprietary models, Claude Fable 5 and GPT 5.6 Sol,\u201d the new open source model \u201cdemonstrated frontier-level performance across our evaluation suite, consistently outperforming other tested models.\u201d Independent analyses from Arena.ai<\/a> and Vals AI<\/a> also suggested that Kimi is competitive with flagship frontier models.<\/p>\n The announcement, which coincided with a speech from Chinese president Xi Jinping<\/a> at the World AI Conference in Shanghai, seems to have spooked Wall Street, with the Nasdaq dropping about 1%<\/a> on Friday as investors sold off stocks in chip companies like Nvidia.<\/p>\n Many of the resulting posts from tech industry figures will sound familiar to those who remember the debate after another Chinese company, DeepSeek, released its open source R1 model in January 2025<\/a>. Except now, everything seems heightened after the Trump administration\u2019s tariff war with China<\/a>, repeated fights over the national security threat supposedly posed by Anthropic<\/a>, and as major AI companies prepare to finally go public<\/a>.<\/p>\n For example, David Sacks \u2014 the Trump administration\u2019s former AI czar and now co-chair of the President\u2019s Council of Advisors on Science and Technology \u2014 contrasted Kimi\u2019s progress<\/a> with a United States that is \u201ctying itself in knots: politicians and bureaucrats are banning new data centers, piling on state regulations, and pushing for new federal agencies to pre-approve frontier models. This is how you lose the AI race.\u201d (The news also gave him an excuse to take a dig<\/a> at Anthropic, calling Claude an example of \u201cwoke lobotomized models.\u201d)<\/p>\n And former Uber CEO Travis Kalanick echoed complaints<\/a> that Chinese are \u201cdistilling off\u201d (i.e., being trained on the outputs of<\/a>) American AI models.<\/p>\n \u201cIf distillation isn\u2019t enforced against, then everyone should be able to distill from everyone else.. otherwise one arm [would be] tied behind American models\u2019 backs,\u201d Kalanick wrote. (Of course, American models have also been built on top of Chinese ones, specifically Kimi<\/a>.)<\/p>\n Meanwhile, OpenAI\u2019s head of strategic futures Dean Ball said<\/a> that Kimi is \u201ca very good model\u201d whose performance probably can\u2019t be \u201cexplained away by distillation or anything like that,\u201d adding that he\u2019s \u201cpersonally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks.\u201d<\/p>\n In fact, Ball suggested that \u201cprobable outcome of an open-weight-model-dominant world is full AI communism,\u201d where AI is treated as \u201ca \u2018public good\u2019 which will ultimately be provided by the state as a kind of \u2018digital public infrastructure.\u2019\u201d<\/p>\n \u201cThis future strikes me as a dystopian hellscape, but I\u2019ve never met an open-weight models advocate who doesn\u2019t ultimately concede this is where things end,\u201d said Ball. He even suggested that the Trump administration (which he used to work for<\/a>) will eventually realize it needs to \u201ccreate large amounts of regulatory risk around the use of open-weight Chinese models.\u201d<\/p>\n \u201cYou don\u2019t need to \u2018ban open source\u2019 (one of the dumber motifs of AI policy discussion),\u201d Ball said. \u201cYou just need to direct every agency to issue soft law that creates FUD [fear, uncertainty, and doubt]. \u2018A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models.\u2019 It needn\u2019t be that well justified. You just create enough regulatory risk that every regulated enterprise backs off.\u201d<\/p>\n However, Shakeel Hashim, editor of the AI-focused publication Transformer, argued that much of the worry is overblown<\/a>, both because Kimi \u201clikely does not have dangerous cyber capabilities,\u201d and because the Chinese government will face \u201cextremely similar incentives\u201d to restrict open Chinese models once they develop those capabilities.<\/p>\n<\/div>\n When you purchase through links in our articles, we may earn a small commission<\/a>. This doesn\u2019t affect our editorial independence.<\/em><\/p>\n