swyx@swyx
51AI 编辑部评分,满分 100
2026-08-12 15:12· 18分钟前
AI 导读

swyx 称一篇关于利用 API 漏洞提取前沿模型隐藏推理的论文已成为今年最重要研究之一。该论文方法学解释不清,swyx 提供了笔记与进一步蒸馏。引用推文显示,研究者已通过所有前沿 AI 公司 API 的漏洞提取隐藏推理,并验证推理 token 数与计费思考 token 1:1 匹配。

this is already one of the most important papers of this year.

https://www.latent.space/p/ainews-how-to-steal-a-reasoning-trace

the methodology doesnt seem clearly explained so here are some notes with a further distillation

Alexander PanfilovWe can finally talk about it: We found a way to extract hidden reasoning of frontier models using a vulnerability in the APIs of every frontier AI company. We v...

来源:swyx · x.com

swyx · @swyx · X·2026-08-12 15:12·18分钟前
AI 导读

swyx 称一篇关于利用 API 漏洞提取前沿模型隐藏推理的论文已成为今年最重要研究之一。该论文方法学解释不清,swyx 提供了笔记与进一步蒸馏。引用推文显示,研究者已通过所有前沿 AI 公司 API 的漏洞提取隐藏推理,并验证推理 token 数与计费思考 token 1:1 匹配。

this is already one of the most important papers of this year.

https://www.latent.space/p/ainews-how-to-steal-a-reasoning-trace

the methodology doesnt seem clearly explained so here are some notes with a further distillation

Alexander PanfilovWe can finally talk about it: We found a way to extract hidden reasoning of frontier models using a vulnerability in the APIs of every frontier AI company. We v...

来源:swyx· x.com