自动化检测大模型漏洞，用多评委验证真实失败。
AdversaBench: Automated LLM Red-Teaming with Multi-Judge Confirmation and Cross-Model Transferability
- 五种结构化操作变异提示，生成难例。
- 45个种子全产生可确认的失败，指令遵循类最难。
- 攻击样本可跨模型迁移，揭示通用行为弱点。
大规模评估大语言模型的对抗性漏洞需要高效生成难题输入并可靠确认失败。我们提出AdversaBench，一个端到端红队测试流水线：通过五种结构化算子变异初始提示，查询目标模型，并由三名评委加一名元评委进行失败确认。在3类共45个种子（推理、指令遵循、工具使用）上实验，所有种子均产生确认失败。四个关键发现：第一，算子效果因类别差异显著——inject_distractor在指令遵循种子上平均奖励为0.00，而在推理和工具使用上达0.80-0.83；第二，二元失败率掩盖难度：指令遵循类平均需2.4次攻击迭代，其他类别仅1.1次，生存曲线清晰显示差距；第三，评委两两一致率达80-87%，但因标签偏斜导致科恩卡帕系数接近零，需以类别层面分歧率评估；第四，针对Llama 3.1 8B生成的对抗提示可零样本迁移至Llama 3.3 70B，表明突变策略捕捉的是普遍行为模式而非模型特异性弱点。代码、数据集与分析脚本见https://github.com/khanak0509/AdversaBench。
原文摘要 · Abstract (English)
Scaling adversarial evaluation of large language models requires both a method for generating hard inputs and a reliable way to confirm that resulting failures are real. We present AdversaBench, an end-to-end red-teaming pipeline that mutates seed prompts with five structured operators, queries a target model, and confirms failures through a three-judge panel with a meta-judge tiebreaker. We report experiments on 45 seeds across three categories: reasoning, instruction-following, and tool use. Every seed produced a confirmed failure. Four findings stand out. First, operator effectiveness varies sharply by category: inject_distractor scores 0.00 mean reward on instruction-following seeds but 0.80-0.83 on reasoning and tool-use. Second, binary failure rate hides difficulty: instruction-following seeds required 2.4 attacker iterations on average versus 1.1 for other categories, a gap visible in survival curves. Third, pairwise judge agreement of 80-87% coexists with near-zero Cohen's kappa due to label skew; category-level disagreement rates are more informative. Fourth, adversarial prompts generated against Llama 3.1 8B transfer zero-shot to Llama 3.3 70B, suggesting the mutations exploit general behavioral patterns rather than model-specific weaknesses. Code, dataset, and analysis scripts are available at https://github.com/khanak0509/AdversaBench .
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。