九原客 (@9hills) “利用 Claude 推理块加密跨模型可复用的漏洞，让 Haiku 恢复 Opus 的加密思维链。 有点” — TopicDigg
九原客
@9hills
加入 February 2010
0 正在关注 0 粉丝
利用 Claude 推理块加密跨模型可复用的漏洞，让 Haiku 恢复 Opus 的加密思维链。
有点意思，不过并不能证明完美恢复，类似于提示词越狱。
We 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 verified that our reasoning token count matches billed API thinking tokens 1:1 for most of the prompts we queried.
显示更多
