Moonshot AI faces mounting US scrutiny over Kimi K3 distillation allegations

Moonshot AI, the Chinese company behind the Kimi chatbot, has become one of the most closely watched names in artificial intelligence. Anthropic has accused the firm of siphoning outputs from its Claude models. White House officials have alleged similar tactics involving another Anthropic model. OpenAI, for its part, has acknowledged the strength of Moonshot’s latest system, Kimi K3, without directly accusing the company of copying its work. The allegations, one by one Start with Anthropic. On September 10, 2026, the company accused Moonshot of pushing approximately 300,000 Kimi user requests through at least 5,380 fraudulent proxy accounts based in Singapore and Japan. According to Anthropic, those requests were primarily aimed at Claude Opus, its flagship model. The alleged goal was to pull data out of Claude and feed it back into Moonshot’s own systems. The US government has made its own claims. In July 2026, officials including OSTP Director Michael Kratsios and Treasury Secretary Scott Bessent said Moonshot allegedly distilled data from Anthropic’s Fable model to build Kimi K3. The officials claimed the company used sophisticated methods to avoid detection. They also alleged M...