You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
公開ウェイトの出自と、利用者に知らせない転送についてご意見ください / Open-weight provenance and undisclosed relaying: your views welcome
#4
Anthropicが2026年9月10日に公開した脅威レポート(レポートページ / PDF 154ページ)は、Kimiの開発元であるMoonshot AIが、利用者のリクエストをKimiで処理せずClaudeへ転送し、Claudeの回答をKimiの回答として表示していたと報告しています(GTG-16002)。DeepSeek・Alibaba・Zhipu・Xiaomiにも別の指摘があります。
Anthropic's September 10, 2026 threat report (page / 154-page PDF) states that Moonshot AI, the maker of Kimi, silently forwarded customer requests to Claude and displayed Claude's responses to its users (GTG-16002). DeepSeek, Alibaba, Zhipu and Xiaomi are named for related activity.
Our article keeps two things apart:
The report concerns Moonshot's service operations, not a finding about the contents of the open Kimi K2.6 weights
So it is not a finding that data from Sakana Chat / Sakana Namazu users, which run on those open weights, was relayed anywhere
That leaves questions we cannot settle ourselves, and we would value your views.
How far should we ask about the provenance of open weights? The weights are distributed; what they learned from is not.
What should we make of relaying user requests to another company's model without telling them? The report cites requests to analyze CCTV footage and live credentials that ended up with a third party.
If you see an official statement from Moonshot or a rebuttal from any named company, please post it here. As of September 11, 2026 we could not find one.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
記事「「国産AI」の土台をもう一段下まで辿ったら、外から確かめられない層に突き当たった」に関する情報提供・ご意見のお願いです。
Anthropicが2026年9月10日に公開した脅威レポート(レポートページ / PDF 154ページ)は、Kimiの開発元であるMoonshot AIが、利用者のリクエストをKimiで処理せずClaudeへ転送し、Claudeの回答をKimiの回答として表示していたと報告しています(GTG-16002)。DeepSeek・Alibaba・Zhipu・Xiaomiにも別の指摘があります。
記事では次の2点を分けて書きました。
そのうえで、こちらでは結論を出せない論点が残りました。ご意見をいただけると助かります。
誤りのご指摘も歓迎します。
English
This thread is for the article "We Traced a 'Domestic AI' One Layer Deeper and Hit a Layer Nobody Outside Can Verify".
Anthropic's September 10, 2026 threat report (page / 154-page PDF) states that Moonshot AI, the maker of Kimi, silently forwarded customer requests to Claude and displayed Claude's responses to its users (GTG-16002). DeepSeek, Alibaba, Zhipu and Xiaomi are named for related activity.
Our article keeps two things apart:
That leaves questions we cannot settle ourselves, and we would value your views.
Corrections are welcome too.
All reactions