Chinese contentPenumbra:针对监管义务的样本高效对抗搜索方法
这篇论文做了个叫 Penumbra 的合规测试工具,能用极少的生成次数找出刚踩到监管红线的回复对,金融和医疗场景挺实用。
这篇论文做了个叫 Penumbra 的合规测试工具,能用极少的生成次数找出刚踩到监管红线的回复对,金融和医疗场景挺实用。
For developers, this matters because it shows that current LLM-based automated program repair tools can be tricked into fixing bugs in a way that introduces new security vulnerabilities.