Michael Kratsios, director of the White House Office of Science and Technology Policy, named Moonshot AI on Wednesday. His claim: the administration has information that Kimi K3 was built with large-scale distillation of Anthropic’s Fable, run through a purpose-built internal platform that could switch between access methods to avoid detection.
What Kimi K3 is
An open-weight LLM, not a chatbot app. 2.8 trillion parameters, sparse MoE with 16 of 896 experts active, 1M-token context, native vision, tuned for long-horizon coding and agent work. It shipped July 16; full weights drop July 27. That date is the point — a frontier-class model anyone can download, allegedly trained on a US lab’s outputs.
Why this one lands differently
Anthropic already said in February it traced 3.4 million Claude exchanges to fraudulent accounts linked to senior Moonshot employees. That was one company accusing another. A government saying it out loud is new, and it follows the Qwen allegations by months.
Moonshot hasn’t responded. Beijing called the February findings groundless. HN put 474 comments on it in a day.
Distillation was the open secret behind how fast the gap closed. It’s now a trade issue, and the enforcement lever isn’t a lawsuit — it’s who gets an API key.
You Might Also Like
- Anthropic Accuses Alibabas Qwen lab of the Largest Ever Distillation Attack 28 8m Claude Queries
- Kimi cli Hits 9600 Github Stars as Moonshot ai Walks Into Claude Codes Territory
- Anthropic Just Launched Code Review in Claude Code and 54 of prs now get Real Feedback
- Claude Replay Turns Your Anthropic Claude Code Sessions Into Shareable Video Like Replays
- 13b Into Openai yet Microsoft Copilot Cowork Runs on Anthropic Claude
Top AI Product 
Leave a comment