Chinese Model Kimi K3 Sparks US AI Policy and IP Fights
Jul 18 - Jul 20
Moonshot AI's open-weight model beat US benchmarks cheaply, prompting warnings from a White House adviser and distillation accusations.
Friday 7/18, David Sacks, the venture investor serving as the White House's AI and crypto policy lead, warned the US is on track to lose the AI race to China after Moonshot AI's Kimi K3 climbed to the top of a leading coding benchmark, beating Anthropic's flagship Fable 5. The 2.8-trillion-parameter model posted an Elo score of 1679 on Arena.ai's Frontend Code leaderboard, with full weights set for public release by July 27. Sacks argued the US is "tying itself in knots" with data-center bans and a push for federal pre-approval of model releases, calling that restrictive domestic policy a bigger threat than China's progress itself.
Saturday 7/19, Moonshot AI confirmed the July 27 open-weight release date, under a modified MIT license, with a 1-million-token context window, native image input, and pricing of $3 per million input tokens and $15 per million output. Independent testing from Artificial Analysis gave the model an Intelligence Index score of 57, calling it a "sweet spot" of high performance and low cost against leading US models - in one test it beat Claude Fable 5 and GPT-5.6 Sol on game design visuals and pacing for just $0.03. Vercel CEO Guillermo Rauch, who ran internal stealth evals, called it "top-tier at cybersecurity" with "raw IQ," pushing back on chatter that Moonshot had overfit its published benchmarks.
By Sunday 7/20, Kimi K3's rise had reignited debate over AI model distillation - querying US labs' APIs to generate training data for rival systems. Anthropic had accused DeepSeek, Moonshot AI and MiniMax in February 2026 of doing this at massive scale using thousands of fake accounts. US figures including Uber co-founder Travis Kalanick called for stronger intellectual-property rules to protect American labs' advantage, while others in the industry argued distillation is standard practice that drives progress and that restricting it could slow innovation broadly.