arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2610.17820cs.CRcs.AI
收藏

相同攻击,不同评判者:Jev、LLM评判者与专用安全分类器的首次受控对抗性比较

Same Attack, Different Judge: A First Controlled Adversarial Comparison of Jev, LLM Judges, and Purpose-Built Safety Classifiers

Nankai University(南开大学);Zhongguancun Laboratory(中关村实验室)

Xiang Li, Aichun Huang, Mingming Zhang, Zheli Liu

AI总结:

本文首次受控对比Jev、LLM评判者及专用安全分类器的对抗性表现,发现Jev在规则类任务中鲁棒性强但判断类任务中表现弱,且速度最快、误报率最高。

摘要:

TypeSafe AI的Jev是一种新型安全评判器,它不生成文本和标签,仅回答开发者预先定义的问题并返回每个答案的概率。该厂商承认对抗性输入可改变其答案,但未提供相关数据。本文对Jev开展了首次受控对抗性评估:构建测试工具,向各系统发送相同载荷并遵循相同策略,在4项公开数据任务(从基于规则的输入审核、注入检测、路由,到基于判断的钓鱼检测)上运行13个家族的50种攻击变体;评估对象包括Jev、5个GPT与GLM评判者,以及2个本地分类器(Llama Guard 3、Prompt Guard 2)。研究发现,Jev在输入审核任务中最具鲁棒性:8种结构化输入伪造体未绕过其任何决策,仅询问“为何有人可能提出有害请求”的语句绕过了其9%的拦截,而GPT-6 Astra的该指标为56%;但在钓鱼检测任务中,相同伪造体绕过了其标记邮件的19%-29%,且无攻击时Jev仅检测到62%的钓鱼邮件,而GLM评判者的该指标为77%-81%。Jev也是速度最快的评判者,但标记了最多的良性请求(27%);本地分类器的失效方式不同:训练示例语句使Llama Guard 3漏检了其曾标记的62%钓鱼内容。这些结果表明,Jev在具有明确检查规则的任务中可抵御攻击,但在需要整体判断的任务中更易受影响。

英文摘要:

TypeSafe AI's Jev is a new kind of safety judge. It generates no text and no label: it answers questions the developer defines in advance and returns a probability for each answer. The vendor admits that adversarial input can change these answers but gives no numbers. In this paper, we present the first controlled adversarial evaluation of Jev. We build a test harness that sends every system the same payload under the same policy, and run it on four public-data tasks, from rule-based ones (input moderation, injection detection, routing) to a judgment-based one (phishing detection), with fifty attack variants from thirteen families. The targets are Jev, five GPT and GLM judges, and two local classifiers (Llama Guard 3, Prompt Guard 2). We find that Jev is the most robust system on input moderation: eight forgeries of its structured input bypass none of its decisions, and a sentence asking only why someone might make a harmful request bypasses 9% of its blocks, against 56% for GPT-6 Astra. On phishing detection. However, the same forgeries bypass 19-29% of the emails it flags, and even without attack Jev detects only 62% of phishing emails, against 77-81% for the GLM judges. Jev is also the fastest judge, but it flags the most benign requests (27%). The local classifiers fail differently: a training-example statement makes Llama Guard 3 miss 62% of the phishing it had flagged. These results suggest that Jev resists attacks on tasks with a clear rule to check but is more exposed on tasks that call for an overall judgment.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑