研究者称发现可窃取推理轨迹的漏洞并已被修补
AI 导读
研究者 Joachim Schaeffer 表示，过去几周他在研究一项名为 Stealing Reasoning Traces 的新工作，即利用漏洞窃取模型的推理轨迹。他提到可能并非只有他们利用了该漏洞，目前这些漏洞已被修补，更多细节将在后续推文中说明。
正文
Over the last weeks I’ve been working on something new: Stealing Reasoning Traces.
Unfortunately, it might be that we’ve not been the only ones who managed to exploit this vulnerability. Fortunately, these vulnerabilities are patched now.
More in the 🧵