
OpenAI 9 月 30 日发布安全披露,把一轮提取隐藏推理的行动核心集群指向月之暗面:7 月峰值两天 1.6 万次请求;这篇逐段读完那份披露,把观测与归属分开摆。

| 时间 | 谁指控 | 指控的手法 | 关键数字 | 出处 |
|---|---|---|---|---|
| 2026-02 | OpenAI 致函美国国会议员 | 说 DeepSeek 用不断变化的手法,从 OpenAI 等前沿实验室的模型里提取结果 | 未给出提取次数 | Bloomberg |
| 2026-02-23 | Anthropic 公开报告 | 说 DeepSeek、月之暗面、MiniMax 用输出蒸馏 Claude 的能力 | 逾 1,600 万次交互、约 2.4 万个欺诈账号;其中月之暗面逾 340 万次 | Anthropic |
| 2026-09-10 | Anthropic 9 月威胁情报报告 | 说月之暗面把用户请求转路由到 Claude Opus,并用跨会话重放提取思维链 | 一次监测的 10 天内近 30 万条请求、5,380 个虚假账号 | Anthropic |
| 2026-09-30 | OpenAI 安全披露 | 说有人跨对话重放加密推理,让较弱的模型逐字转写 | 7 月 24—25 日峰值 16,000 次尝试、4,000 多个用户;相关集群逾 1.5 万个用户 | OpenAI |
"We recently identified and disrupted a coordinated campaign designed to extract protected reasoning from our models, with the earliest observed activity occurring in the first week of July."
"It is unclear whether all operators we observed during the relevant time period originated from a single actor. However, we attribute a core cluster of the activity to individuals associated with Moonshot AI, the developer of Kimi."
"The operators did not break our encryption, compromise a database, or gain direct access to stored user conversations."
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.